fig2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8225b1ea-700e-40ef-809b-633c6d57cfe7",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "## Data Preprocess"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ff8ac009-70e6-4b89-843d-fa852386b92e",
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "import os\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = '7'\n",
    "from deepsramp import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "27bace17-20f4-4d53-908c-f2d0e10ebd78",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/nfs/fanrui/work/deepsramp/deepsramp/preprocess.py:29: DtypeWarning: Columns (0) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  gtf = pd.read_csv(filename, sep='\\t', comment='#', header=None)\n",
      "51696870it [02:25, 356522.98it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1min 38s, sys: 1min 22s, total: 3min 1s\n",
      "Wall time: 3min 3s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "gtf = read_gtf('data/Homo_sapiens.GRCh37.87.gtf.gz')\n",
    "seqs = read_seq('data/Homo_sapiens.GRCh37.dna.primary_assembly.fa.gz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "394db153-b0f7-4cd7-a0f3-9c9627dd3352",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 14.7 s, sys: 852 ms, total: 15.6 s\n",
      "Wall time: 15.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sdf = get_mature_splice(gtf, seqs, grp=['id', 'trans'], keepmax=False)\n",
    "ndf = get_mature_cds(gtf, sdf, grp=['id', 'trans'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "bd4196a0-25f0-4c56-b1bb-b11b2c357ca9",
   "metadata": {
    "jupyter": {
     "source_hidden": true
    }
   },
   "outputs": [],
   "source": [
    "# used for random \n",
    "# import random\n",
    "\n",
    "# def read_m6a_bed(file, name, ndf, pat, get_pos):\n",
    "#     m6adf = pd.read_csv(file, sep='\\t', header=None)\n",
    "#     m6adf = m6adf[m6adf[0].apply(len) < 6]\n",
    "#     m6adf[0] = m6adf[0].str.slice(3).replace({'M': 'MT'})\n",
    "#     tqdm.pandas(desc=name)\n",
    "#     m6adf['pos'] = m6adf.progress_apply(lambda x: [\n",
    "#         [\n",
    "#             gene, \n",
    "#             trans, \n",
    "#             (lambda y: y if y != -1 and len(re.findall(pat, i.seq[y-2:y+3])) else -1)(get_pos(x[1], i)),\n",
    "#             # i.length # for max length\n",
    "#         ] for (gene, trans), i in get_trans(x[0], x[5], x[1], ndf).iterrows()\n",
    "#     ], axis=1)\n",
    "#     m6adf['pos'] = m6adf.pos.apply(lambda x: [i for i in x if i[2] != -1]) # drop -1\n",
    "#     # m6adf['pos'] = m6adf.pos.apply(lambda x: [max(x, key=lambda y: y[3])[:3]] if x else []) # keep max length\n",
    "#     m6adf['pos'] = m6adf.pos.apply(lambda x: [random.choice(x)] if x else []) # random\n",
    "#     res = pd.DataFrame(sum(m6adf.pos, []), columns=ndf.index.names+[name])\n",
    "#     res = res[res[name] != -1]\n",
    "#     res = res.groupby(ndf.index.names).agg(lambda x: set(x))\n",
    "#     return res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "7bfd493c-804b-4fe0-9b8d-924fc6f7907f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "molm3: 100%|██████████| 2050/2050 [00:11<00:00, 175.24it/s]]\n",
      "molm2: 100%|██████████| 2866/2866 [00:16<00:00, 175.54it/s]]\n",
      "molm1: 100%|██████████| 8113/8113 [00:45<00:00, 176.78it/s]]\n",
      "abcam: 100%|██████████| 9536/9536 [00:52<00:00, 182.54it/s]]\n",
      "sysy: 100%|██████████| 12051/12051 [01:05<00:00, 183.55it/s]]\n",
      "cd8t: 100%|██████████| 19677/19677 [01:49<00:00, 180.16it/s]]\n",
      "a549: 100%|██████████| 23478/23478 [02:07<00:00, 183.57it/s]]\n",
      "hela1: 100%|██████████| 37603/37603 [03:22<00:00, 186.06it/s]\n",
      "hela4: 100%|██████████| 37187/37187 [03:25<00:00, 180.93it/s]\n",
      "hela2: 100%|██████████| 43407/43407 [03:50<00:00, 188.13it/s]\n",
      "hela3: 100%|██████████| 46350/46350 [04:11<00:00, 184.64it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 9 s, sys: 9.35 s, total: 18.4 s\n",
      "Wall time: 4min 32s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "files = ['data/GSE/GSE63753/GSE63753_hek293.abcam.CIMS.m6A.9536.bed',\n",
    "         'data/GSE/GSE63753/GSE63753_hek293.sysy.CITS.m6A.12051.bed',\n",
    "         'data/GSE/GSE71154/GSM1828594_precise_m6A_sites_CD8T_hg19.bed',\n",
    "         'data/GSE/GSE71154/GSM1828596_precise_m6A_sites_A549_hg19.bed',\n",
    "         'data/GSE/GSE98623/GSM2602070_molm13_rep1.tag.uniq.sites.bed.gz',\n",
    "         'data/GSE/GSE98623/GSM2602071_molm13_rep2.tag.uniq.sites.bed.gz',\n",
    "         'data/GSE/GSE98623/GSM2602072_molm13_rep3.tag.uniq.sites.bed.gz',\n",
    "         'data/GSE/GSE86336/GSM2300426_precise_m6A_sites_HeLa_CA_hg19.bed.gz',\n",
    "         'data/GSE/GSE86336/GSM2300427_precise_m6A_sites_HeLa_NU_hg19.bed.gz',\n",
    "         'data/GSE/GSE86336/GSM2300428_precise_m6A_sites_HeLa_CY_hg19.bed.gz',\n",
    "         'data/GSE/GSE86336/GSM2300429_precise_m6A_sites_HeLa_CYpolyA_hg19.bed.gz',\n",
    "        ]\n",
    "names = ['abcam', 'sysy', 'cd8t', 'a549', 'molm1', 'molm2', 'molm3', 'hela1', 'hela2', 'hela3', 'hela4']\n",
    "\n",
    "res = read_m6a_beds(files, names, ndf, RRACH_PAT, get_mature_pos)\n",
    "res = res.dropna(subset=['m6a_pos'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f6d32fc6-c1f7-4aef-939a-a7cb8fa75bff",
   "metadata": {},
   "outputs": [],
   "source": [
    "utils.save(res, 'data/GSE/mature_rrach.data')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "666d5139-6204-42df-8a79-69fb3b1f780b",
   "metadata": {},
   "outputs": [],
   "source": [
    "utils.save(res, 'data/GSE/mature_rrach_random.data')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3646360-4dfc-42e5-af38-a189648c401d",
   "metadata": {},
   "source": [
    "### YTHDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "68a28c4b-c24e-488a-8354-2daa20c6041d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/nfs/fanrui/work/deepsramp/deepsramp/preprocess.py:28: DtypeWarning: Columns (0) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  gtf = pd.read_csv(filename, sep='\\t', comment='#', header=None)\n",
      "51662809it [02:17, 376715.85it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1min 39s, sys: 1min 18s, total: 2min 57s\n",
      "Wall time: 2min 54s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "gtf38 = read_gtf('data/Homo_sapiens.GRCh38.79.gtf.gz')\n",
    "seqs38 = read_seq('data/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "3875baed-9dc4-46e0-8731-a1332671cee5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 30 s, sys: 2.56 s, total: 32.5 s\n",
      "Wall time: 32.5 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sdf38 = get_mature_splice(gtf38, seqs38, grp=['id', 'trans'], keepmax=False, biotype=None)\n",
    "ndf38 = get_mature_cds(gtf38, sdf38, grp=['id', 'trans'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "34a08953-7048-48b8-9b46-38bd383e9b8b",
   "metadata": {},
   "outputs": [],
   "source": [
    "yth = pd.read_csv('data/sramp1/YTHDF_mature.csv', usecols=[0,1,2])\n",
    "yth['Position'] -= 1\n",
    "\n",
    "ythpos = yth[yth.Class == 1].groupby('TranscriptID').agg({'Position': set})\n",
    "ythpos.columns = ['ythdf_test']\n",
    "ythneg = yth[yth.Class == -1].groupby('TranscriptID').agg({'Position': set})\n",
    "ythneg.columns = ['ythdf_test_neg']\n",
    "ythdf = ythpos.merge(ythneg, left_index=True, right_index=True, how='outer')\n",
    "\n",
    "ythdf['ythdf_test'] = ythdf['ythdf_test'].apply(lambda x: utils.notnaor(x, set()))\n",
    "ythdf['ythdf_test_neg'] = ythdf['ythdf_test_neg'].apply(lambda x: utils.notnaor(x, set()))\n",
    "\n",
    "ythdf = ythdf.merge(ndf38.reset_index(), left_index=True, right_on='trans').reset_index().set_index('trans')\n",
    "utils.save(ythdf, 'data/GSE/compare_mature_ythdf_test.data')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "40372430-eedf-4ce6-b2b6-fb02c6386c0b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# df = utils.load('compare_mature.data')\n",
    "# df['ythdf_train'] = df['m6a_pos'].copy()\n",
    "# df['_train_len'] = (df['ythdf_train'].apply(len) * 1).apply(round)\n",
    "# df['ythdf_train_neg'] = df.apply(lambda x: set(random.choices(list(x.m6a_neg), k=x._train_len)) if len(x.m6a_neg) else set(), axis=1)\n",
    "# # df = df.merge(ythdf, left_index=True, right_index=True)\n",
    "# utils.save(df, 'data/GSE/compare_mature.data')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4a9f7ba-d845-4d2f-a539-23b6a9458ed8",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "## Sample Negtive Samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9e53cb04-5472-4fff-8105-697e1ceceb6e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "def get_m6a_neg_ratio4(osdf, label='m6a_pos', train_ratio=1, test_ratio=10, pat=RRACH_PAT):\n",
    "    sdf = osdf\n",
    "    \n",
    "    train_label = f'{label}_train'\n",
    "    test_label = f'{label}_test'\n",
    "    \n",
    "    sdf['_train_len'] = (sdf[train_label].apply(len) * train_ratio).apply(round)\n",
    "    sdf['_test_len'] = (sdf[test_label].apply(len) * test_ratio).apply(round)\n",
    "    \n",
    "    sdf['_neg'] = sdf.apply(lambda x: (random.choices(list(x.m6a_neg), k=x._train_len + x._test_len)) if len(x.m6a_neg) else [], axis=1)\n",
    "    \n",
    "    sdf[f'{label}_train_neg'] = sdf.apply(lambda x: set(x._neg[x._test_len:]), axis=1)\n",
    "    sdf[f'{label}_test_neg'] = sdf.apply(lambda x: set(x._neg[:x._test_len]) - x[f'{label}_train_neg'], axis=1)\n",
    "    \n",
    "    return sdf\n",
    "\n",
    "def get_m6a_pos(sdf, test_label='m6a_pos'):\n",
    "    if test_label != 'm6a_pos':\n",
    "        pos = sdf.apply(lambda x: set(sum([list(utils.notnaor(x[i], [])) for i in datasets if i != test_label], [])), axis=1)\n",
    "        sdf[f'{test_label}_train'] = pos\n",
    "        sdf[f'{test_label}_test'] = sdf.apply(lambda x: utils.notnaor(x[test_label], set()) - x[f'{test_label}_train'], axis=1)\n",
    "    else:\n",
    "        pos = sdf.apply(lambda x: set(sum([list(utils.notnaor(x[i], [])) for i in datasets if i != test_label], [])), axis=1)\n",
    "        sdf['m6a_pos'] = pos\n",
    "        \n",
    "    return sdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "90f0a8f3-bbe4-43e8-a30d-36016eac0953",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 6/6 [00:04<00:00,  1.36it/s]\n"
     ]
    }
   ],
   "source": [
    "datasets = ['a549', 'cd8t', 'hela', 'sysy', 'abcam', 'molm', 'ythdf'][:-1]\n",
    "\n",
    "res = utils.load('data/GSE/mature_rrach.data')\n",
    "df = res.reset_index().set_index('trans')\n",
    "df['molm'] = df.apply(lambda x: set().union(*[i for i in [x.molm1, x.molm2, x.molm3] if i == i]), axis=1)\n",
    "df['hela'] = df.apply(lambda x: set().union(*[i for i in [x.hela1, x.hela2, x.hela3, x.hela4] if i == i]), axis=1)\n",
    "df = get_m6a_pos(df) # all pos -> m6a_pos\n",
    "df = get_m6a_neg(df, pat=RRACH_PAT) # all RRACH -> m6a_neg\n",
    "\n",
    "for i in tqdm(datasets):\n",
    "    df = get_m6a_pos(df, test_label=i) # not-i_pos -> i_train; i_pos - m6a_pos -> i_test\n",
    "    df = get_m6a_neg_ratio4(df, label=i, test_ratio=10, train_ratio=1) # m6a_neg * ratio -> i_test_neg, i_train_neg\n",
    "\n",
    "# for YTHDF\n",
    "df['ythdf_train'] = df['m6a_pos'].copy()\n",
    "df['_train_len'] = (df['ythdf_train'].apply(len) * 1).apply(round)\n",
    "df['ythdf_train_neg'] = df.apply(lambda x: set(random.choices(list(x.m6a_neg), k=x._train_len)) if len(x.m6a_neg) else set(), axis=1)\n",
    "# df = df.merge(ythdf, left_index=True, right_index=True)\n",
    "\n",
    "utils.save(df, 'data/GSE/compare_mature.data')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "2d12b74c-95d9-4b9e-80d7-85383dbc9ea4",
   "metadata": {
    "jupyter": {
     "source_hidden": true
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 6/6 [00:04<00:00,  1.22it/s]\n"
     ]
    }
   ],
   "source": [
    "# used for random\n",
    "# datasets = ['a549', 'cd8t', 'hela', 'sysy', 'abcam', 'molm', 'ythdf'][:-1]\n",
    "\n",
    "# res = utils.load('data/GSE/mature_rrach_random.data')\n",
    "# df = res.reset_index().set_index('trans')\n",
    "# df['molm'] = df.apply(lambda x: set().union(*[i for i in [x.molm1, x.molm2, x.molm3] if i == i]), axis=1)\n",
    "# df['hela'] = df.apply(lambda x: set().union(*[i for i in [x.hela1, x.hela2, x.hela3, x.hela4] if i == i]), axis=1)\n",
    "# df = get_m6a_pos(df) # all pos -> m6a_pos\n",
    "# df = get_m6a_neg(df, pat=RRACH_PAT) # all RRACH -> m6a_neg\n",
    "\n",
    "# for i in tqdm(datasets):\n",
    "#     df = get_m6a_pos(df, test_label=i) # not-i_pos -> i_train; i_pos - m6a_pos -> i_test\n",
    "#     df = get_m6a_neg_ratio4(df, label=i, test_ratio=10, train_ratio=1) # m6a_neg * ratio -> i_test_neg, i_train_neg\n",
    "\n",
    "# # for YTHDF\n",
    "# df['ythdf_train'] = df['m6a_pos'].copy()\n",
    "# df['_train_len'] = (df['ythdf_train'].apply(len) * 1).apply(round)\n",
    "# df['ythdf_train_neg'] = df.apply(lambda x: set(random.choices(list(x.m6a_neg), k=x._train_len)) if len(x.m6a_neg) else set(), axis=1)\n",
    "# # df = df.merge(ythdf, left_index=True, right_index=True)\n",
    "\n",
    "# utils.save(df, 'data/GSE/compare_mature_random.data')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f31c1c70-8da7-4ed7-9fe3-487a3aa2e750",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "## Train Single-deepSRAMP and DeepPromise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "29c9a880-33fb-4380-a63f-1d0b1ac16e41",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = utils.load('data/GSE/compare_mature.data')\n",
    "ythdf = utils.load('data/GSE/compare_mature_ythdf_test.data')\n",
    "datasets = ['abcam', 'sysy', 'cd8t', 'a549', 'molm', 'hela', 'ythdf']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "c70af562-f675-4ffe-be76-f748508f0a19",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = utils.load('data/GSE/compare_mature_random.data')\n",
    "ythdf = utils.load('data/GSE/compare_mature_ythdf_test.data')\n",
    "datasets = ['abcam', 'sysy', 'cd8t', 'a549', 'molm', 'hela', 'ythdf']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "7f622f1b-c4e3-49c9-b6bf-8ef37a0569e6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "abcam 50345 45508 2564 18117\n",
      "sysy 51203 46298 1706 12608\n",
      "cd8t 48302 43634 4607 31463\n",
      "a549 47915 43289 4994 35687\n",
      "molm 48670 43995 4239 29320\n",
      "hela 36747 34110 16162 95274\n",
      "ythdf 52909 47778 5318 53180\n"
     ]
    }
   ],
   "source": [
    "for i in (datasets[:-1]):\n",
    "    print(i,\n",
    "          df[f'{i}_train'].apply(len).sum(),\n",
    "          df[f'{i}_train_neg'].apply(len).sum(),\n",
    "          df[f'{i}_test'].apply(len).sum(),\n",
    "          df[f'{i}_test_neg'].apply(len).sum(),\n",
    "    )\n",
    "\n",
    "print('ythdf',\n",
    "     df[f'ythdf_train'].apply(len).sum(),\n",
    "     df[f'ythdf_train_neg'].apply(len).sum(),\n",
    "     ythdf[f'ythdf_test'].apply(len).sum(),\n",
    "     ythdf[f'ythdf_test_neg'].apply(len).sum(),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a59c1a18-b0c5-4b5b-9bf6-a32ae12ea661",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "abcam\t1.1062890041311417\t0.14152453496715792\t0\t0\t0\t0\n",
      "sysy\t1.1059441012570737\t0.13531091370558376\t0\t0\t0\t0\n",
      "cd8t\t1.1069807947930512\t0.14642596065219465\t0\t0\t0\t0\n",
      "a549\t1.106863175402527\t0.13993891332978395\t0\t0\t0\t0\n",
      "molm\t1.1062620752358223\t0.14457708049113233\t0\t0\t0\t0\n",
      "hela\t1.0773087071240106\t0.16963704683334382\t0\t0\t0\t0\n"
     ]
    }
   ],
   "source": [
    "for i in (datasets[:-1]):\n",
    "    rtrain = df[f'{i}_train'].apply(len).sum() / df[f'{i}_train_neg'].apply(len).sum()\n",
    "    rtest = df[f'{i}_test'].apply(len).sum() / df[f'{i}_test_neg'].apply(len).sum()\n",
    "    overlap_pos = df.apply(lambda x: len(x[f'{i}_train'] & x[f'{i}_test']), axis=1).sum()\n",
    "    overlap_neg = df.apply(lambda x: len(x[f'{i}_train_neg'] & x[f'{i}_test_neg']), axis=1).sum()\n",
    "    overlap_pn = df.apply(lambda x: len(x[f'{i}_train'] & x[f'{i}_train_neg']), axis=1).sum()\n",
    "    overlap_tpn = df.apply(lambda x: len(x[f'{i}_test'] & x[f'{i}_test_neg']), axis=1).sum()\n",
    "    print(i, rtrain, rtest, overlap_pos, overlap_neg, overlap_pn, overlap_tpn, sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "eee528bf-ff91-4e6d-8016-14c92c0ace27",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_res = []\n",
    "for name in datasets:\n",
    "    i = name\n",
    "    print(f'Generating test data - [{name}]')\n",
    "    if name == 'ythdf':\n",
    "        testds = df2ds(ythdf, downsample=False, pos_label=f'{i}_test', neg_label=f'{i}_test_neg')\n",
    "    else:\n",
    "        testds = df2ds(df, downsample=False, pos_label=f'{i}_test', neg_label=f'{i}_test_neg')\n",
    "    testdl = ds2dl(testds, drop_last=False, num_workers=0)\n",
    "\n",
    "    print(f'Generating train data - [{name}]')\n",
    "    trainds = df2ds(df, downsample=False, pos_label=f'{i}_train', neg_label=f'{i}_train_neg')\n",
    "    traindl = ds2dl(trainds, drop_last=True, num_workers=8)\n",
    "    \n",
    "    print(f'Building model - [{name}]')\n",
    "    device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "    model = SRAMP().to(device)\n",
    "    loss_fn = TriLoss()\n",
    "    optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)\n",
    "    scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer, 1, 2)\n",
    "\n",
    "    print(f'Start to train - [{name}]')\n",
    "    train(model, loss_fn, optimizer, scheduler, traindl, testdl, epochs=1, device=device)\n",
    "    test_res += [test_loop(testdl, model, device, loss_fn)]\n",
    "\n",
    "    dps = []\n",
    "    for i in ['onehot', 'enac', 'embedding', 'ensemble']:\n",
    "        print(i)\n",
    "        traindl = ds2dl(trainds, drop_last=True, num_workers=8)\n",
    "        if i == 'ensemble':\n",
    "            dp = DeepPromiseEnsemble(*dps).to(device)\n",
    "        else:\n",
    "            dp = DeepPromise(i).to(device)\n",
    "        loss_fn = DPLoss()\n",
    "        optimizer = torch.optim.Adam(dp.parameters(), lr=1e-3)\n",
    "        scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer, 1, 2)\n",
    "        print(f'Start to train - [{i}]')\n",
    "        train(dp, loss_fn, optimizer, scheduler, traindl, testdl, epochs=1, device=device)\n",
    "        test_res += [dp_test_loop(testdl, dp, device, loss_fn)]\n",
    "        dps += [dp]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "1889492c-0aa0-4aa2-b13f-49b2d665169f",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(model, 'model/deepsramp_mature_sramp1data_all.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "e41ecdbc-5e44-46c4-a66d-31b22cf2a14a",
   "metadata": {},
   "outputs": [],
   "source": [
    "utils.save(test_res, 'model/deepsramp_mature_sramp1data_all.metric')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cc58fab2-ad76-411b-aae6-d1632e6e8526",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f0c9d9a1-a966-4f19-9fb6-a7d37d972858",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "d25c0529-5183-42f4-a150-3ac1165d0f91",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "### WHISTLE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "9761b5fe-c6e8-49dd-a25d-75ae5f7054e7",
   "metadata": {},
   "outputs": [],
   "source": [
    "def df2bed(testdf, pos_label, neg_label, name=\"\", mode=\"mature\"):\n",
    "    if name == \"\": name = pos_label\n",
    "    whi_test = []\n",
    "    for _, i in tqdm(testdf.iterrows(), total=testdf.shape[0], desc=pos_label):\n",
    "        for j in utils.notnaor(i[pos_label], []):\n",
    "            genome = get_genome_pos(j, i, mode)\n",
    "            whi_test += [[f'chr{i[0] if i[0] != \"MT\" else \"M\"}', genome, genome+1, i[6], 1]]\n",
    "        for j in utils.notnaor(i[neg_label], []):\n",
    "            genome = get_genome_pos(j, i, mode)\n",
    "            whi_test += [[f'chr{i[0] if i[0] != \"MT\" else \"M\"}', genome, genome+1, i[6], 0]]\n",
    "\n",
    "    pd.DataFrame(whi_test, columns=['#chromsome', 'modStart', 'modEnd', 'strand', 'label']).to_csv(f'whistle/traindata1/{name.replace(' ', '.')}_whistle.bed', sep='\\t', index=None, header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "e449d404-14eb-44fd-8cb2-ea045e40876e",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "abcam_test: 100%|██████████| 10331/10331 [00:00<00:00, 10633.44it/s]\n",
      "abcam_train: 100%|██████████| 10331/10331 [00:02<00:00, 3955.75it/s]\n",
      "sysy_test: 100%|██████████| 10331/10331 [00:00<00:00, 12232.71it/s]\n",
      "sysy_train: 100%|██████████| 10331/10331 [00:02<00:00, 4354.19it/s]\n",
      "cd8t_test: 100%|██████████| 10331/10331 [00:01<00:00, 8572.22it/s]\n",
      "cd8t_train: 100%|██████████| 10331/10331 [00:03<00:00, 3129.92it/s]\n",
      "a549_test: 100%|██████████| 10331/10331 [00:00<00:00, 10575.24it/s]\n",
      "a549_train: 100%|██████████| 10331/10331 [00:01<00:00, 5295.57it/s]\n",
      "molm_test: 100%|██████████| 10331/10331 [00:01<00:00, 8485.21it/s]\n",
      "molm_train: 100%|██████████| 10331/10331 [00:02<00:00, 4567.42it/s]\n",
      "hela_test: 100%|██████████| 10331/10331 [00:02<00:00, 4012.36it/s]\n",
      "hela_train: 100%|██████████| 10331/10331 [00:03<00:00, 3422.69it/s]\n",
      "ythdf_test: 100%|██████████| 2797/2797 [00:01<00:00, 2100.00it/s]\n",
      "ythdf_train: 100%|██████████| 10331/10331 [00:02<00:00, 4107.01it/s]\n"
     ]
    }
   ],
   "source": [
    "for i in datasets:\n",
    "    if i == 'ythdf':\n",
    "        df2bed(ythdf, f'{i}_test', f'{i}_test_neg')\n",
    "    else:\n",
    "        df2bed(df, f'{i}_test', f'{i}_test_neg')\n",
    "    df2bed(df, f'{i}_train', f'{i}_train_neg')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4bb07fb0-89fe-4c69-bba6-03caf48aef95",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "191a215e-2917-4bdc-ae85-63056c55a19d",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "## Train Multi-deepSRAMP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "8622f913-8bdf-40ff-8a57-c43f8114baa2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:20<00:00, 495.21it/s]\n",
      "100%|██████████| 10331/10331 [00:19<00:00, 517.80it/s]\n",
      "100%|██████████| 10331/10331 [00:22<00:00, 459.50it/s]\n",
      "100%|██████████| 10331/10331 [00:20<00:00, 515.44it/s]\n",
      "100%|██████████| 10331/10331 [00:15<00:00, 653.01it/s]\n",
      "100%|██████████| 10331/10331 [00:20<00:00, 509.86it/s]\n",
      "100%|██████████| 10331/10331 [00:18<00:00, 560.81it/s]\n",
      "100%|██████████| 10331/10331 [00:22<00:00, 468.00it/s]\n",
      "100%|██████████| 10331/10331 [00:01<00:00, 7223.05it/s]\n",
      "100%|██████████| 10331/10331 [00:17<00:00, 578.53it/s]\n",
      "100%|██████████| 10331/10331 [00:14<00:00, 733.97it/s]\n",
      "100%|██████████| 10331/10331 [00:19<00:00, 541.19it/s]\n",
      "100%|██████████| 10331/10331 [00:02<00:00, 4557.05it/s]\n",
      "100%|██████████| 10331/10331 [00:01<00:00, 9939.86it/s]\n",
      "100%|██████████| 10331/10331 [00:18<00:00, 573.60it/s]\n",
      "100%|██████████| 10331/10331 [00:07<00:00, 1364.57it/s]\n",
      "100%|██████████| 10331/10331 [00:17<00:00, 585.44it/s]\n",
      "100%|██████████| 10331/10331 [00:02<00:00, 4304.79it/s]\n",
      "100%|██████████| 10331/10331 [00:05<00:00, 1951.19it/s]\n",
      "100%|██████████| 10331/10331 [00:06<00:00, 1495.82it/s]\n",
      "100%|██████████| 10331/10331 [00:02<00:00, 4763.38it/s]\n",
      "100%|██████████| 10331/10331 [00:19<00:00, 520.08it/s]\n",
      "100%|██████████| 10331/10331 [00:14<00:00, 737.77it/s]\n",
      "100%|██████████| 2797/2797 [00:03<00:00, 899.59it/s]]]\n",
      "100%|██████████| 10331/10331 [00:12<00:00, 845.28it/s] \n",
      "100%|██████████| 10331/10331 [00:15<00:00, 686.20it/s]\n",
      "100%|██████████| 10331/10331 [00:39<00:00, 264.81it/s]\n",
      "100%|██████████| 2797/2797 [00:30<00:00, 92.01it/s] \n"
     ]
    }
   ],
   "source": [
    "data_dfs = {}\n",
    "datasets = ['abcam', 'sysy', 'cd8t', 'a549', 'molm', 'hela', 'ythdf']\n",
    "\n",
    "def task_transform(name):\n",
    "    dfs = []\n",
    "    for train in ['train', 'test']:\n",
    "        if name == 'ythdf' and train == 'test':\n",
    "            idf = ythdf\n",
    "            indf = ndf38\n",
    "        else:\n",
    "            idf = df\n",
    "            indf = ndf\n",
    "        tdf = []\n",
    "        exist_grp = set()\n",
    "        for neg in ['', '_neg']:\n",
    "            for _, i in tqdm(idf.iterrows(), total=idf.shape[0]):\n",
    "                for j in i[f'{name}_{train}{neg}']:\n",
    "                    genome_pos = get_genome_pos(j, i)\n",
    "                    if i[0] + i[6] + str(genome_pos) in exist_grp: continue\n",
    "                    else: exist_grp.add(i[0] + i[6] + str(genome_pos))\n",
    "                    for trans, k in indf.loc[i.id].iterrows():\n",
    "                        mature_pos = get_mature_pos(genome_pos, k)\n",
    "                        if mature_pos != -1:\n",
    "                            tdf += [(i.id, trans, mature_pos, i[0] + i[6] + str(genome_pos), neg)]\n",
    "        tdf = pd.DataFrame(tdf, columns=['id', 'trans', 'pos', 'grp', 'label'])\n",
    "        tdf['label'] = (tdf['label'] == '') * 1\n",
    "        tdf = tdf.merge(indf.reset_index(), left_on=['id', 'trans'], right_on=['id', 'trans'])\n",
    "        dfs += [tdf]\n",
    "    return name, dfs\n",
    "\n",
    "df = utils.load('data/GSE/compare_mature.data')\n",
    "ythdf = utils.load('data/GSE/compare_mature_ythdf_test.data')\n",
    "dfs = Parallel(n_jobs=8)(delayed(task_transform)(name) for name in datasets)\n",
    "data_dfs = dict(dfs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "f3587126-56c6-46d9-92d9-2fdd8fd82540",
   "metadata": {},
   "outputs": [],
   "source": [
    "utils.save(data_dfs, 'data/GSE/compare_mature_multi.data')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "3f1f12d1-0ee1-48be-b149-5bde41603e38",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dfs = utils.load('data/GSE/compare_mature_multi.data')\n",
    "datasets = ['abcam', 'sysy', 'cd8t', 'a549', 'molm', 'hela', 'ythdf']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f952a479-c2d9-41dd-92df-928441923263",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_res = []\n",
    "for i in datasets:\n",
    "    traindf, testdf = data_dfs[i]\n",
    "    \n",
    "    trainds = df2ds_multi(traindf)\n",
    "    traindl = ds2dl(trainds, batch_size=128, drop_last=True, num_workers=4)\n",
    "    \n",
    "    testds = df2ds_multi(testdf)\n",
    "    testdl = ds2dl(testds, batch_size=128, shuffle=False, num_workers=4)\n",
    "    \n",
    "    device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "    utils.setup_seed(42)\n",
    "    \n",
    "    model = MultiSRAMP().to(device)\n",
    "    loss_fn = TriLoss()\n",
    "    optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)\n",
    "    scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer, 1, 2)\n",
    "    \n",
    "    print(f'Start to train - [{i}]')\n",
    "    train(model, loss_fn, optimizer, scheduler, traindl, testdl, epochs=31, device=device)\n",
    "    test_res += [test_loop(testdl, model, device, loss_fn)]\n",
    "\n",
    "    torch.save(model, f'model/full_400_{i}.model')\n",
    "    utils.save(test_res, f'model/full_400_{i}.metrics')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a3e25d6f-6508-4f18-8dc5-a42a926a7f54",
   "metadata": {
    "jp-MarkdownHeadingCollapsed": true
   },
   "source": [
    "## Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "7b815b13-6d23-4125-88b7-713d2b6ca523",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from matplotlib import font_manager\n",
    "font_manager.fontManager.addfont('/home/fanrui/fonts/ARIAL.TTF')\n",
    "plt.rcParams['svg.fonttype'] = 'none'\n",
    "plt.rcParams['font.family'] = 'Arial'\n",
    "plt.rcParams['font.size'] = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "dfda3d31-219f-4cc2-9046-7532f649adba",
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets_1 = ['a549', 'cd8t', 'hela', 'abcam', 'sysy', 'molm', 'ythdf']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "12e40e8a-e012-4b84-8412-bb62fe13abda",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>MultiSRAMP</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a549</th>\n",
       "      <td>0.573559</td>\n",
       "      <td>0.908690</td>\n",
       "      <td>0.703239</td>\n",
       "      <td>0.905091</td>\n",
       "      <td>0.905535</td>\n",
       "      <td>0.674819</td>\n",
       "      <td>0.823551</td>\n",
       "      <td>0.969817</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cd8t</th>\n",
       "      <td>0.549798</td>\n",
       "      <td>0.914261</td>\n",
       "      <td>0.686664</td>\n",
       "      <td>0.889984</td>\n",
       "      <td>0.893095</td>\n",
       "      <td>0.656464</td>\n",
       "      <td>0.784318</td>\n",
       "      <td>0.962912</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hela</th>\n",
       "      <td>0.558173</td>\n",
       "      <td>0.875387</td>\n",
       "      <td>0.681683</td>\n",
       "      <td>0.881645</td>\n",
       "      <td>0.880732</td>\n",
       "      <td>0.636167</td>\n",
       "      <td>0.778131</td>\n",
       "      <td>0.953155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abcam</th>\n",
       "      <td>0.658184</td>\n",
       "      <td>0.817083</td>\n",
       "      <td>0.729076</td>\n",
       "      <td>0.939880</td>\n",
       "      <td>0.924641</td>\n",
       "      <td>0.691309</td>\n",
       "      <td>0.848362</td>\n",
       "      <td>0.971667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sysy</th>\n",
       "      <td>0.655909</td>\n",
       "      <td>0.917351</td>\n",
       "      <td>0.764907</td>\n",
       "      <td>0.934826</td>\n",
       "      <td>0.932741</td>\n",
       "      <td>0.740856</td>\n",
       "      <td>0.867039</td>\n",
       "      <td>0.980306</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>molm</th>\n",
       "      <td>0.586938</td>\n",
       "      <td>0.797122</td>\n",
       "      <td>0.676070</td>\n",
       "      <td>0.918712</td>\n",
       "      <td>0.903323</td>\n",
       "      <td>0.630836</td>\n",
       "      <td>0.784457</td>\n",
       "      <td>0.956643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ythdf</th>\n",
       "      <td>0.506625</td>\n",
       "      <td>0.866997</td>\n",
       "      <td>0.639539</td>\n",
       "      <td>0.916544</td>\n",
       "      <td>0.912087</td>\n",
       "      <td>0.621194</td>\n",
       "      <td>0.745976</td>\n",
       "      <td>0.967138</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            Precision    Recall        F1  Specificity  Accuracy       MCC  \\\n",
       "MultiSRAMP                                                                   \n",
       "a549         0.573559  0.908690  0.703239     0.905091  0.905535  0.674819   \n",
       "cd8t         0.549798  0.914261  0.686664     0.889984  0.893095  0.656464   \n",
       "hela         0.558173  0.875387  0.681683     0.881645  0.880732  0.636167   \n",
       "abcam        0.658184  0.817083  0.729076     0.939880  0.924641  0.691309   \n",
       "sysy         0.655909  0.917351  0.764907     0.934826  0.932741  0.740856   \n",
       "molm         0.586938  0.797122  0.676070     0.918712  0.903323  0.630836   \n",
       "ythdf        0.506625  0.866997  0.639539     0.916544  0.912087  0.621194   \n",
       "\n",
       "               AUPRC     AUROC  \n",
       "MultiSRAMP                      \n",
       "a549        0.823551  0.969817  \n",
       "cd8t        0.784318  0.962912  \n",
       "hela        0.778131  0.953155  \n",
       "abcam       0.848362  0.971667  \n",
       "sysy        0.867039  0.980306  \n",
       "molm        0.784457  0.956643  \n",
       "ythdf       0.745976  0.967138  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_res = []\n",
    "for i in datasets_1:\n",
    "    test_res += utils.load(f'model/full_400_{i}.metrics')\n",
    "\n",
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets_1])\n",
    "t.index.name = 'MultiSRAMP'\n",
    "t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "376bf7a7-0399-440a-a556-11000df83bd2",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "703d8ada-23f7-4c8f-9879-6056c2440970",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "35bca9ec-1590-4c6a-856f-a04dec6a781f",
   "metadata": {},
   "outputs": [],
   "source": [
    "single = utils.load('test_res')\n",
    "whistle = utils.load('whistle_performance.pk')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "29e205d1-fe84-43a6-b2a7-b8a79edad1c8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "      <th>method</th>\n",
       "      <th>dataset</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.573559</td>\n",
       "      <td>0.908690</td>\n",
       "      <td>0.703239</td>\n",
       "      <td>0.905091</td>\n",
       "      <td>0.905535</td>\n",
       "      <td>0.674819</td>\n",
       "      <td>0.823551</td>\n",
       "      <td>0.969817</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.549798</td>\n",
       "      <td>0.914261</td>\n",
       "      <td>0.686664</td>\n",
       "      <td>0.889984</td>\n",
       "      <td>0.893095</td>\n",
       "      <td>0.656464</td>\n",
       "      <td>0.784318</td>\n",
       "      <td>0.962912</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.558173</td>\n",
       "      <td>0.875387</td>\n",
       "      <td>0.681683</td>\n",
       "      <td>0.881645</td>\n",
       "      <td>0.880732</td>\n",
       "      <td>0.636167</td>\n",
       "      <td>0.778131</td>\n",
       "      <td>0.953155</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.658184</td>\n",
       "      <td>0.817083</td>\n",
       "      <td>0.729076</td>\n",
       "      <td>0.939880</td>\n",
       "      <td>0.924641</td>\n",
       "      <td>0.691309</td>\n",
       "      <td>0.848362</td>\n",
       "      <td>0.971667</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.655909</td>\n",
       "      <td>0.917351</td>\n",
       "      <td>0.764907</td>\n",
       "      <td>0.934826</td>\n",
       "      <td>0.932741</td>\n",
       "      <td>0.740856</td>\n",
       "      <td>0.867039</td>\n",
       "      <td>0.980306</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.586938</td>\n",
       "      <td>0.797122</td>\n",
       "      <td>0.676070</td>\n",
       "      <td>0.918712</td>\n",
       "      <td>0.903323</td>\n",
       "      <td>0.630836</td>\n",
       "      <td>0.784457</td>\n",
       "      <td>0.956643</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.506625</td>\n",
       "      <td>0.866997</td>\n",
       "      <td>0.639539</td>\n",
       "      <td>0.916544</td>\n",
       "      <td>0.912087</td>\n",
       "      <td>0.621194</td>\n",
       "      <td>0.745976</td>\n",
       "      <td>0.967138</td>\n",
       "      <td>deepSRAMP</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.546170</td>\n",
       "      <td>0.916700</td>\n",
       "      <td>0.684510</td>\n",
       "      <td>0.893407</td>\n",
       "      <td>0.896266</td>\n",
       "      <td>0.657284</td>\n",
       "      <td>0.809242</td>\n",
       "      <td>0.967068</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.281342</td>\n",
       "      <td>0.807970</td>\n",
       "      <td>0.417356</td>\n",
       "      <td>0.711183</td>\n",
       "      <td>0.723065</td>\n",
       "      <td>0.356591</td>\n",
       "      <td>0.408472</td>\n",
       "      <td>0.836564</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.285945</td>\n",
       "      <td>0.813977</td>\n",
       "      <td>0.423217</td>\n",
       "      <td>0.715555</td>\n",
       "      <td>0.727637</td>\n",
       "      <td>0.364457</td>\n",
       "      <td>0.416247</td>\n",
       "      <td>0.839664</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.280265</td>\n",
       "      <td>0.803364</td>\n",
       "      <td>0.415558</td>\n",
       "      <td>0.711295</td>\n",
       "      <td>0.722598</td>\n",
       "      <td>0.353656</td>\n",
       "      <td>0.406990</td>\n",
       "      <td>0.835500</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.287591</td>\n",
       "      <td>0.819584</td>\n",
       "      <td>0.425778</td>\n",
       "      <td>0.715891</td>\n",
       "      <td>0.728620</td>\n",
       "      <td>0.368452</td>\n",
       "      <td>0.421138</td>\n",
       "      <td>0.843148</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0.525889</td>\n",
       "      <td>0.901671</td>\n",
       "      <td>0.664321</td>\n",
       "      <td>0.880971</td>\n",
       "      <td>0.883615</td>\n",
       "      <td>0.631663</td>\n",
       "      <td>0.762971</td>\n",
       "      <td>0.957303</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0.286466</td>\n",
       "      <td>0.815064</td>\n",
       "      <td>0.423935</td>\n",
       "      <td>0.702730</td>\n",
       "      <td>0.717078</td>\n",
       "      <td>0.359330</td>\n",
       "      <td>0.423336</td>\n",
       "      <td>0.837370</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0.287777</td>\n",
       "      <td>0.796180</td>\n",
       "      <td>0.422751</td>\n",
       "      <td>0.711471</td>\n",
       "      <td>0.722290</td>\n",
       "      <td>0.354475</td>\n",
       "      <td>0.411020</td>\n",
       "      <td>0.830566</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0.285106</td>\n",
       "      <td>0.814413</td>\n",
       "      <td>0.422356</td>\n",
       "      <td>0.700982</td>\n",
       "      <td>0.715470</td>\n",
       "      <td>0.357362</td>\n",
       "      <td>0.420542</td>\n",
       "      <td>0.836991</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>0.290201</td>\n",
       "      <td>0.820273</td>\n",
       "      <td>0.428725</td>\n",
       "      <td>0.706226</td>\n",
       "      <td>0.720793</td>\n",
       "      <td>0.365891</td>\n",
       "      <td>0.430604</td>\n",
       "      <td>0.841318</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>0.550000</td>\n",
       "      <td>0.830343</td>\n",
       "      <td>0.661703</td>\n",
       "      <td>0.884753</td>\n",
       "      <td>0.876862</td>\n",
       "      <td>0.608912</td>\n",
       "      <td>0.719610</td>\n",
       "      <td>0.940900</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>0.313293</td>\n",
       "      <td>0.667739</td>\n",
       "      <td>0.426485</td>\n",
       "      <td>0.751716</td>\n",
       "      <td>0.739537</td>\n",
       "      <td>0.319617</td>\n",
       "      <td>0.386854</td>\n",
       "      <td>0.789229</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>0.311650</td>\n",
       "      <td>0.657778</td>\n",
       "      <td>0.422922</td>\n",
       "      <td>0.753542</td>\n",
       "      <td>0.739653</td>\n",
       "      <td>0.314271</td>\n",
       "      <td>0.382662</td>\n",
       "      <td>0.785262</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>0.310370</td>\n",
       "      <td>0.675164</td>\n",
       "      <td>0.425253</td>\n",
       "      <td>0.745513</td>\n",
       "      <td>0.735310</td>\n",
       "      <td>0.318766</td>\n",
       "      <td>0.386666</td>\n",
       "      <td>0.788323</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>0.314258</td>\n",
       "      <td>0.666316</td>\n",
       "      <td>0.427087</td>\n",
       "      <td>0.753353</td>\n",
       "      <td>0.740730</td>\n",
       "      <td>0.320243</td>\n",
       "      <td>0.388349</td>\n",
       "      <td>0.789862</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>0.630769</td>\n",
       "      <td>0.865182</td>\n",
       "      <td>0.729609</td>\n",
       "      <td>0.931472</td>\n",
       "      <td>0.923571</td>\n",
       "      <td>0.697999</td>\n",
       "      <td>0.800973</td>\n",
       "      <td>0.970933</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>0.249629</td>\n",
       "      <td>0.690504</td>\n",
       "      <td>0.366693</td>\n",
       "      <td>0.719147</td>\n",
       "      <td>0.715733</td>\n",
       "      <td>0.282344</td>\n",
       "      <td>0.305007</td>\n",
       "      <td>0.781301</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>0.252008</td>\n",
       "      <td>0.680539</td>\n",
       "      <td>0.367812</td>\n",
       "      <td>0.726681</td>\n",
       "      <td>0.721182</td>\n",
       "      <td>0.282417</td>\n",
       "      <td>0.307345</td>\n",
       "      <td>0.780807</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>0.253447</td>\n",
       "      <td>0.678781</td>\n",
       "      <td>0.369084</td>\n",
       "      <td>0.729457</td>\n",
       "      <td>0.723418</td>\n",
       "      <td>0.283743</td>\n",
       "      <td>0.300196</td>\n",
       "      <td>0.780999</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>0.250862</td>\n",
       "      <td>0.682298</td>\n",
       "      <td>0.366845</td>\n",
       "      <td>0.724302</td>\n",
       "      <td>0.719296</td>\n",
       "      <td>0.281461</td>\n",
       "      <td>0.312620</td>\n",
       "      <td>0.784659</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>0.636654</td>\n",
       "      <td>0.700468</td>\n",
       "      <td>0.667038</td>\n",
       "      <td>0.943423</td>\n",
       "      <td>0.913302</td>\n",
       "      <td>0.618262</td>\n",
       "      <td>0.754266</td>\n",
       "      <td>0.954203</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>0.230793</td>\n",
       "      <td>0.584633</td>\n",
       "      <td>0.330942</td>\n",
       "      <td>0.724237</td>\n",
       "      <td>0.706929</td>\n",
       "      <td>0.219310</td>\n",
       "      <td>0.272439</td>\n",
       "      <td>0.721408</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>0.232069</td>\n",
       "      <td>0.594384</td>\n",
       "      <td>0.333808</td>\n",
       "      <td>0.721643</td>\n",
       "      <td>0.705865</td>\n",
       "      <td>0.223726</td>\n",
       "      <td>0.279324</td>\n",
       "      <td>0.722217</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>0.229418</td>\n",
       "      <td>0.585803</td>\n",
       "      <td>0.329711</td>\n",
       "      <td>0.721532</td>\n",
       "      <td>0.704705</td>\n",
       "      <td>0.217752</td>\n",
       "      <td>0.274474</td>\n",
       "      <td>0.717274</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>0.234389</td>\n",
       "      <td>0.594384</td>\n",
       "      <td>0.336201</td>\n",
       "      <td>0.725230</td>\n",
       "      <td>0.709008</td>\n",
       "      <td>0.226873</td>\n",
       "      <td>0.280136</td>\n",
       "      <td>0.724931</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>0.555813</td>\n",
       "      <td>0.735315</td>\n",
       "      <td>0.633086</td>\n",
       "      <td>0.915041</td>\n",
       "      <td>0.892339</td>\n",
       "      <td>0.579111</td>\n",
       "      <td>0.724957</td>\n",
       "      <td>0.941736</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.577731</td>\n",
       "      <td>0.331327</td>\n",
       "      <td>0.723909</td>\n",
       "      <td>0.705444</td>\n",
       "      <td>0.215870</td>\n",
       "      <td>0.264006</td>\n",
       "      <td>0.712047</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>0.232209</td>\n",
       "      <td>0.564992</td>\n",
       "      <td>0.329142</td>\n",
       "      <td>0.729911</td>\n",
       "      <td>0.709080</td>\n",
       "      <td>0.212332</td>\n",
       "      <td>0.262367</td>\n",
       "      <td>0.710649</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>0.227573</td>\n",
       "      <td>0.568530</td>\n",
       "      <td>0.325039</td>\n",
       "      <td>0.721010</td>\n",
       "      <td>0.701749</td>\n",
       "      <td>0.206968</td>\n",
       "      <td>0.262433</td>\n",
       "      <td>0.710155</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>0.233869</td>\n",
       "      <td>0.573720</td>\n",
       "      <td>0.332286</td>\n",
       "      <td>0.728274</td>\n",
       "      <td>0.708752</td>\n",
       "      <td>0.216944</td>\n",
       "      <td>0.266217</td>\n",
       "      <td>0.713845</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>0.495843</td>\n",
       "      <td>0.766758</td>\n",
       "      <td>0.602236</td>\n",
       "      <td>0.918907</td>\n",
       "      <td>0.904572</td>\n",
       "      <td>0.567755</td>\n",
       "      <td>0.672568</td>\n",
       "      <td>0.953009</td>\n",
       "      <td>deepSRAMP_Single</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>0.176292</td>\n",
       "      <td>0.681129</td>\n",
       "      <td>0.280091</td>\n",
       "      <td>0.668967</td>\n",
       "      <td>0.670113</td>\n",
       "      <td>0.212559</td>\n",
       "      <td>0.227489</td>\n",
       "      <td>0.737963</td>\n",
       "      <td>DP-oh</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>0.168394</td>\n",
       "      <td>0.697429</td>\n",
       "      <td>0.271286</td>\n",
       "      <td>0.641745</td>\n",
       "      <td>0.646991</td>\n",
       "      <td>0.203123</td>\n",
       "      <td>0.226826</td>\n",
       "      <td>0.731311</td>\n",
       "      <td>DP-enac</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>0.184023</td>\n",
       "      <td>0.663682</td>\n",
       "      <td>0.288149</td>\n",
       "      <td>0.693897</td>\n",
       "      <td>0.691050</td>\n",
       "      <td>0.220546</td>\n",
       "      <td>0.230595</td>\n",
       "      <td>0.741036</td>\n",
       "      <td>DP-emb</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>0.177947</td>\n",
       "      <td>0.677227</td>\n",
       "      <td>0.281838</td>\n",
       "      <td>0.674579</td>\n",
       "      <td>0.674828</td>\n",
       "      <td>0.214297</td>\n",
       "      <td>0.228247</td>\n",
       "      <td>0.739045</td>\n",
       "      <td>DeepPromise</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>0.270101</td>\n",
       "      <td>0.861033</td>\n",
       "      <td>0.411208</td>\n",
       "      <td>0.674391</td>\n",
       "      <td>0.697303</td>\n",
       "      <td>0.360016</td>\n",
       "      <td>0.432978</td>\n",
       "      <td>0.848884</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>A549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>0.275522</td>\n",
       "      <td>0.848274</td>\n",
       "      <td>0.415944</td>\n",
       "      <td>0.673394</td>\n",
       "      <td>0.695731</td>\n",
       "      <td>0.356469</td>\n",
       "      <td>0.409995</td>\n",
       "      <td>0.837547</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>CD8T</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>0.332228</td>\n",
       "      <td>0.706596</td>\n",
       "      <td>0.451955</td>\n",
       "      <td>0.759074</td>\n",
       "      <td>0.751463</td>\n",
       "      <td>0.355040</td>\n",
       "      <td>0.411530</td>\n",
       "      <td>0.800272</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>HeLa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>0.270802</td>\n",
       "      <td>0.854631</td>\n",
       "      <td>0.411283</td>\n",
       "      <td>0.688610</td>\n",
       "      <td>0.708397</td>\n",
       "      <td>0.363352</td>\n",
       "      <td>0.381290</td>\n",
       "      <td>0.840961</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>HEK293 Abcam</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>0.264330</td>\n",
       "      <td>0.597114</td>\n",
       "      <td>0.366443</td>\n",
       "      <td>0.764807</td>\n",
       "      <td>0.744016</td>\n",
       "      <td>0.265625</td>\n",
       "      <td>0.295377</td>\n",
       "      <td>0.746557</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>HEK293 SYSY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>0.266725</td>\n",
       "      <td>0.648974</td>\n",
       "      <td>0.378066</td>\n",
       "      <td>0.742053</td>\n",
       "      <td>0.730296</td>\n",
       "      <td>0.281541</td>\n",
       "      <td>0.318042</td>\n",
       "      <td>0.757620</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>MOLM13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>0.185513</td>\n",
       "      <td>0.782116</td>\n",
       "      <td>0.299894</td>\n",
       "      <td>0.648118</td>\n",
       "      <td>0.660573</td>\n",
       "      <td>0.255903</td>\n",
       "      <td>0.263162</td>\n",
       "      <td>0.782185</td>\n",
       "      <td>WHISTLE</td>\n",
       "      <td>YTHDF</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Precision    Recall        F1  Specificity  Accuracy       MCC     AUPRC  \\\n",
       "0    0.573559  0.908690  0.703239     0.905091  0.905535  0.674819  0.823551   \n",
       "1    0.549798  0.914261  0.686664     0.889984  0.893095  0.656464  0.784318   \n",
       "2    0.558173  0.875387  0.681683     0.881645  0.880732  0.636167  0.778131   \n",
       "3    0.658184  0.817083  0.729076     0.939880  0.924641  0.691309  0.848362   \n",
       "4    0.655909  0.917351  0.764907     0.934826  0.932741  0.740856  0.867039   \n",
       "5    0.586938  0.797122  0.676070     0.918712  0.903323  0.630836  0.784457   \n",
       "6    0.506625  0.866997  0.639539     0.916544  0.912087  0.621194  0.745976   \n",
       "7    0.546170  0.916700  0.684510     0.893407  0.896266  0.657284  0.809242   \n",
       "8    0.281342  0.807970  0.417356     0.711183  0.723065  0.356591  0.408472   \n",
       "9    0.285945  0.813977  0.423217     0.715555  0.727637  0.364457  0.416247   \n",
       "10   0.280265  0.803364  0.415558     0.711295  0.722598  0.353656  0.406990   \n",
       "11   0.287591  0.819584  0.425778     0.715891  0.728620  0.368452  0.421138   \n",
       "12   0.525889  0.901671  0.664321     0.880971  0.883615  0.631663  0.762971   \n",
       "13   0.286466  0.815064  0.423935     0.702730  0.717078  0.359330  0.423336   \n",
       "14   0.287777  0.796180  0.422751     0.711471  0.722290  0.354475  0.411020   \n",
       "15   0.285106  0.814413  0.422356     0.700982  0.715470  0.357362  0.420542   \n",
       "16   0.290201  0.820273  0.428725     0.706226  0.720793  0.365891  0.430604   \n",
       "17   0.550000  0.830343  0.661703     0.884753  0.876862  0.608912  0.719610   \n",
       "18   0.313293  0.667739  0.426485     0.751716  0.739537  0.319617  0.386854   \n",
       "19   0.311650  0.657778  0.422922     0.753542  0.739653  0.314271  0.382662   \n",
       "20   0.310370  0.675164  0.425253     0.745513  0.735310  0.318766  0.386666   \n",
       "21   0.314258  0.666316  0.427087     0.753353  0.740730  0.320243  0.388349   \n",
       "22   0.630769  0.865182  0.729609     0.931472  0.923571  0.697999  0.800973   \n",
       "23   0.249629  0.690504  0.366693     0.719147  0.715733  0.282344  0.305007   \n",
       "24   0.252008  0.680539  0.367812     0.726681  0.721182  0.282417  0.307345   \n",
       "25   0.253447  0.678781  0.369084     0.729457  0.723418  0.283743  0.300196   \n",
       "26   0.250862  0.682298  0.366845     0.724302  0.719296  0.281461  0.312620   \n",
       "27   0.636654  0.700468  0.667038     0.943423  0.913302  0.618262  0.754266   \n",
       "28   0.230793  0.584633  0.330942     0.724237  0.706929  0.219310  0.272439   \n",
       "29   0.232069  0.594384  0.333808     0.721643  0.705865  0.223726  0.279324   \n",
       "30   0.229418  0.585803  0.329711     0.721532  0.704705  0.217752  0.274474   \n",
       "31   0.234389  0.594384  0.336201     0.725230  0.709008  0.226873  0.280136   \n",
       "32   0.555813  0.735315  0.633086     0.915041  0.892339  0.579111  0.724957   \n",
       "33   0.232265  0.577731  0.331327     0.723909  0.705444  0.215870  0.264006   \n",
       "34   0.232209  0.564992  0.329142     0.729911  0.709080  0.212332  0.262367   \n",
       "35   0.227573  0.568530  0.325039     0.721010  0.701749  0.206968  0.262433   \n",
       "36   0.233869  0.573720  0.332286     0.728274  0.708752  0.216944  0.266217   \n",
       "37   0.495843  0.766758  0.602236     0.918907  0.904572  0.567755  0.672568   \n",
       "38   0.176292  0.681129  0.280091     0.668967  0.670113  0.212559  0.227489   \n",
       "39   0.168394  0.697429  0.271286     0.641745  0.646991  0.203123  0.226826   \n",
       "40   0.184023  0.663682  0.288149     0.693897  0.691050  0.220546  0.230595   \n",
       "41   0.177947  0.677227  0.281838     0.674579  0.674828  0.214297  0.228247   \n",
       "42   0.270101  0.861033  0.411208     0.674391  0.697303  0.360016  0.432978   \n",
       "43   0.275522  0.848274  0.415944     0.673394  0.695731  0.356469  0.409995   \n",
       "44   0.332228  0.706596  0.451955     0.759074  0.751463  0.355040  0.411530   \n",
       "45   0.270802  0.854631  0.411283     0.688610  0.708397  0.363352  0.381290   \n",
       "46   0.264330  0.597114  0.366443     0.764807  0.744016  0.265625  0.295377   \n",
       "47   0.266725  0.648974  0.378066     0.742053  0.730296  0.281541  0.318042   \n",
       "48   0.185513  0.782116  0.299894     0.648118  0.660573  0.255903  0.263162   \n",
       "\n",
       "       AUROC            method       dataset  \n",
       "0   0.969817         deepSRAMP          A549  \n",
       "1   0.962912         deepSRAMP          CD8T  \n",
       "2   0.953155         deepSRAMP          HeLa  \n",
       "3   0.971667         deepSRAMP  HEK293 Abcam  \n",
       "4   0.980306         deepSRAMP   HEK293 SYSY  \n",
       "5   0.956643         deepSRAMP        MOLM13  \n",
       "6   0.967138         deepSRAMP         YTHDF  \n",
       "7   0.967068  deepSRAMP_Single          A549  \n",
       "8   0.836564             DP-oh          A549  \n",
       "9   0.839664           DP-enac          A549  \n",
       "10  0.835500            DP-emb          A549  \n",
       "11  0.843148       DeepPromise          A549  \n",
       "12  0.957303  deepSRAMP_Single          CD8T  \n",
       "13  0.837370             DP-oh          CD8T  \n",
       "14  0.830566           DP-enac          CD8T  \n",
       "15  0.836991            DP-emb          CD8T  \n",
       "16  0.841318       DeepPromise          CD8T  \n",
       "17  0.940900  deepSRAMP_Single          HeLa  \n",
       "18  0.789229             DP-oh          HeLa  \n",
       "19  0.785262           DP-enac          HeLa  \n",
       "20  0.788323            DP-emb          HeLa  \n",
       "21  0.789862       DeepPromise          HeLa  \n",
       "22  0.970933  deepSRAMP_Single  HEK293 Abcam  \n",
       "23  0.781301             DP-oh  HEK293 Abcam  \n",
       "24  0.780807           DP-enac  HEK293 Abcam  \n",
       "25  0.780999            DP-emb  HEK293 Abcam  \n",
       "26  0.784659       DeepPromise  HEK293 Abcam  \n",
       "27  0.954203  deepSRAMP_Single   HEK293 SYSY  \n",
       "28  0.721408             DP-oh   HEK293 SYSY  \n",
       "29  0.722217           DP-enac   HEK293 SYSY  \n",
       "30  0.717274            DP-emb   HEK293 SYSY  \n",
       "31  0.724931       DeepPromise   HEK293 SYSY  \n",
       "32  0.941736  deepSRAMP_Single        MOLM13  \n",
       "33  0.712047             DP-oh        MOLM13  \n",
       "34  0.710649           DP-enac        MOLM13  \n",
       "35  0.710155            DP-emb        MOLM13  \n",
       "36  0.713845       DeepPromise        MOLM13  \n",
       "37  0.953009  deepSRAMP_Single         YTHDF  \n",
       "38  0.737963             DP-oh         YTHDF  \n",
       "39  0.731311           DP-enac         YTHDF  \n",
       "40  0.741036            DP-emb         YTHDF  \n",
       "41  0.739045       DeepPromise         YTHDF  \n",
       "42  0.848884           WHISTLE          A549  \n",
       "43  0.837547           WHISTLE          CD8T  \n",
       "44  0.800272           WHISTLE          HeLa  \n",
       "45  0.840961           WHISTLE  HEK293 Abcam  \n",
       "46  0.746557           WHISTLE   HEK293 SYSY  \n",
       "47  0.757620           WHISTLE        MOLM13  \n",
       "48  0.782185           WHISTLE         YTHDF  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multi = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])\n",
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (single)])\n",
    "wdata = np.array([[i[-1]['1']['precision'], i[-1]['1']['recall'], i[-1]['1']['f1-score'], i[-1]['0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (whistle)])\n",
    "data = np.concatenate([multi, data, wdata])\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'])\n",
    "methods = ['deepSRAMP']*7 + [i for j in ['A549', 'CD8T', 'HeLa', 'HEK293 Abcam', 'HEK293 SYSY', 'MOLM13', 'YTHDF'] for i in ['deepSRAMP_Single', 'DP-oh', 'DP-enac', 'DP-emb', 'DeepPromise']] + ['WHISTLE']*7\n",
    "ds = ['A549', 'CD8T', 'HeLa', 'HEK293 Abcam', 'HEK293 SYSY', 'MOLM13', 'YTHDF'] + [j for j in ['A549', 'CD8T', 'HeLa', 'HEK293 Abcam', 'HEK293 SYSY', 'MOLM13', 'YTHDF'] for i in ['deepSRAMP_Single', 'DP-oh', 'DP-enac', 'DP-emb', 'DeepPromise']] + [j for j in ['A549', 'CD8T', 'HeLa', 'HEK293 Abcam', 'HEK293 SYSY', 'MOLM13', 'YTHDF']]\n",
    "t['method'] = methods\n",
    "t['dataset'] = ds\n",
    "t.to_csv('fig/performance.csv')\n",
    "t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4b522b27-f2d7-4630-99db-26fba32418d8",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "02331b29-81f5-4e5d-91a7-016fd53c897e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tt = t.set_index(['method', 'dataset']).stack()[['deepSRAMP', 'deepSRAMP_Single', 'WHISTLE', 'DeepPromise']].reset_index()\n",
    "tt['method'] = tt.method.replace({'deepSRAMP_Single': 'deepSRAMP$_{{single}}$'})\n",
    "tt = tt[tt.level_2 !='AUROC']\n",
    "\n",
    "fig = plt.figure(figsize=(6, 4))\n",
    "ax = fig.subplots()\n",
    "ax.spines['top'].set_visible(False)\n",
    "ax.spines['right'].set_visible(False)\n",
    "sns.barplot(data=tt, x='level_2', y=0, hue='method', hue_order=['DeepPromise', 'WHISTLE', 'deepSRAMP$_{{single}}$', 'deepSRAMP'], palette=['#44757A', '#A5A58D', '#CB997E', '#D44C3C'])\n",
    "plt.legend(loc='upper center', frameon=False, ncol=2, bbox_to_anchor=(0.5, 1.2))\n",
    "plt.ylim((0, 1))\n",
    "plt.xlabel('')\n",
    "plt.ylabel('')\n",
    "plt.tight_layout()\n",
    "plt.savefig('fig/fig2_allmetric.svg')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8644fd6a-994b-4692-852f-b2ded3bb69de",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "6734de9a-75d9-4ed2-975b-37fe1768c71f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in ['AUROC', 'AUPRC']:\n",
    "    tt = t[t.dataset != 'YTHDF'][[i, 'method', 'dataset']].set_index(['method', 'dataset']).unstack()[i]\n",
    "    tt = tt.loc[['deepSRAMP', 'deepSRAMP_Single', 'WHISTLE', 'DeepPromise']]\n",
    "    tt.index = ['deepSRAMP', 'deepSRAMP$_{{single}}$', 'WHISTLE', 'DeepPromise']\n",
    "    tt['Average'] = tt.mean(axis=1)\n",
    "    sns.heatmap(tt.T, cmap='Reds', annot=True, fmt='.3f', square=True)\n",
    "    plt.xticks(rotation=30, horizontalalignment='right', verticalalignment='top')\n",
    "    plt.yticks(rotation=0)\n",
    "    plt.xlabel('')\n",
    "    plt.ylabel('')\n",
    "    plt.tight_layout()\n",
    "    plt.savefig(f'fig/fig2_{i}.svg')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ebc3ce3d-fafa-42f9-b631-80d811fd532d",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "bb2bdf90-38ba-445f-9729-b678424f6bae",
   "metadata": {},
   "outputs": [],
   "source": [
    "colors = ['#44757A', '#A5A58D', '#CB997E', '#D44C3C']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "b102569d-a1c1-4901-8181-aff2b63f9501",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 400x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# roc\n",
    "rocs = [[test_res[-1][3], test_res[-1][4]]] + [[i[3], i[4]] for i in single[-5:]] + [[whistle[-1][3], whistle[-1][4]]]\n",
    "aucs = [test_res[-1][1]] + [i[1] for i in single[-5:]] + [whistle[-1][1]]\n",
    "\n",
    "fig = plt.figure(figsize=(4, 4))\n",
    "ax = fig.subplots()\n",
    "ax.spines['top'].set_visible(False)\n",
    "ax.spines['right'].set_visible(False)\n",
    "\n",
    "# sns.lineplot(x=rocs[0][0], y=rocs[0][1], label='SRAMP2')\n",
    "plt.plot(*rocs[0], label=f'deepSRAMP AUROC={aucs[0]:.3f}', color=colors[3])\n",
    "plt.plot(*rocs[1], label=f'deepSRAMP$_{{single}}$ AUROC={aucs[1]:.3f}', color=colors[2])\n",
    "plt.plot(*rocs[6], label=f'WHISTLE AUROC={aucs[6]:.3f}', color=colors[1])\n",
    "plt.plot(*rocs[5], label=f'DeepPromise AUROC={aucs[5]:.3f}', color=colors[0])\n",
    "plt.xlim(-0.01, 1.01)\n",
    "plt.ylim(-0.01, 1.01)\n",
    "plt.xlabel('False Postive Rate')\n",
    "plt.ylabel('True Postive Rate')\n",
    "plt.title('YTHDF1/2 Binding Sites')\n",
    "plt.legend(frameon=False)\n",
    "plt.tight_layout()\n",
    "plt.savefig('fig/fig2_ythdf_roc.svg')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "84dddf63-696f-4578-8bbb-2f619278c0bb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 400x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# prc\n",
    "rocs = [[test_res[-1][7], test_res[-1][6]]] + [(i[7], i[6]) for i in single[-5:]] + [(whistle[-1][7], whistle[-1][6])]\n",
    "aucs = [test_res[-1][2]] + [i[2] for i in single[-5:]] + [whistle[-1][2]]\n",
    "\n",
    "fig = plt.figure(figsize=(4, 4))\n",
    "ax = fig.subplots()\n",
    "ax.spines['top'].set_visible(False)\n",
    "ax.spines['right'].set_visible(False)\n",
    "\n",
    "plt.plot(*rocs[0], label=f'deepSRAMP AUPRC={aucs[0]:.3f}', color=colors[3])\n",
    "plt.plot(*rocs[1], label=f'deepSRAMP$_{{single}}$ AUPRC={aucs[1]:.3f}', color=colors[2])\n",
    "plt.plot(*rocs[6], label=f'WHISTLE AUPRC={aucs[6]:.3f}', color=colors[1])\n",
    "plt.plot(*rocs[5], label=f'DeepPromise AUPRC={aucs[5]:.3f}', color=colors[0])\n",
    "plt.xlim(-0.01, 1.01)\n",
    "plt.ylim(-0.01, 1.01)\n",
    "plt.ylabel('Precision')\n",
    "plt.xlabel('Recall')\n",
    "plt.legend(loc='lower left', frameon=False, bbox_to_anchor=(0, -0.02))\n",
    "plt.title('YTHDF1/2 Binding Sites')\n",
    "plt.tight_layout()\n",
    "plt.savefig('fig/fig2_ythdf_prc.svg')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7a0be7f3-4196-434a-8c07-e5c3c5d8d264",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "e401b35e-855c-4fde-918a-2ff4d76f8fb0",
   "metadata": {},
   "source": [
    "### Random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "8f759c33-11cc-422b-b597-9e933f755a9d",
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets_1 = ['a549', 'cd8t', 'hela', 'abcam', 'sysy', 'molm', 'ythdf']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "09237bdc-a329-4930-9f0f-f1c836bb64a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "ts = []\n",
    "for a in ['dp', 'single']:\n",
    "    for b in ['', '_random']:\n",
    "        test_res = []\n",
    "        for i in datasets_1:\n",
    "            test_res += [utils.load(f'model/full_400_{i}_{a}{b}.metrics')[-1]]\n",
    "        \n",
    "        data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "        t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets_1])\n",
    "        t.loc['average'] = t.iloc[:-1].mean(axis=0)\n",
    "        ts += [t[['AUPRC', 'AUROC']]]\n",
    "\n",
    "ts = pd.concat(ts, axis=1).T\n",
    "ts['method'] = [f'{a}{b}' for a in ['DeepPromise', 'deepSRAMP$_{single}$'] for b in ['_maxlen', '_random'] for _ in range(2)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "7bfd7733-02d3-4cdd-8b34-5108a9e7e0e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "ts.reset_index().set_index(['method', 'index']).to_csv('fig/random.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "d4eeea2f-398a-4615-bd0c-8b4e71fb0bfd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a549</th>\n",
       "      <th>cd8t</th>\n",
       "      <th>hela</th>\n",
       "      <th>abcam</th>\n",
       "      <th>sysy</th>\n",
       "      <th>molm</th>\n",
       "      <th>ythdf</th>\n",
       "      <th>average</th>\n",
       "      <th>method</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUPRC</th>\n",
       "      <td>0.418955</td>\n",
       "      <td>0.420442</td>\n",
       "      <td>0.390339</td>\n",
       "      <td>0.278911</td>\n",
       "      <td>0.311095</td>\n",
       "      <td>0.264470</td>\n",
       "      <td>0.238700</td>\n",
       "      <td>0.347368</td>\n",
       "      <td>DeepPromise_maxlen</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUROC</th>\n",
       "      <td>0.842155</td>\n",
       "      <td>0.838535</td>\n",
       "      <td>0.791095</td>\n",
       "      <td>0.723788</td>\n",
       "      <td>0.785666</td>\n",
       "      <td>0.713264</td>\n",
       "      <td>0.747836</td>\n",
       "      <td>0.782417</td>\n",
       "      <td>DeepPromise_maxlen</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUPRC</th>\n",
       "      <td>0.425007</td>\n",
       "      <td>0.424914</td>\n",
       "      <td>0.384913</td>\n",
       "      <td>0.272305</td>\n",
       "      <td>0.304974</td>\n",
       "      <td>0.262237</td>\n",
       "      <td>0.237371</td>\n",
       "      <td>0.345725</td>\n",
       "      <td>DeepPromise_random</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUROC</th>\n",
       "      <td>0.843230</td>\n",
       "      <td>0.839872</td>\n",
       "      <td>0.788713</td>\n",
       "      <td>0.720002</td>\n",
       "      <td>0.787811</td>\n",
       "      <td>0.711557</td>\n",
       "      <td>0.746492</td>\n",
       "      <td>0.781864</td>\n",
       "      <td>DeepPromise_random</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUPRC</th>\n",
       "      <td>0.793165</td>\n",
       "      <td>0.758831</td>\n",
       "      <td>0.724289</td>\n",
       "      <td>0.765890</td>\n",
       "      <td>0.825184</td>\n",
       "      <td>0.684066</td>\n",
       "      <td>0.649307</td>\n",
       "      <td>0.758571</td>\n",
       "      <td>deepSRAMP$_{single}$_maxlen</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUROC</th>\n",
       "      <td>0.965697</td>\n",
       "      <td>0.957643</td>\n",
       "      <td>0.941825</td>\n",
       "      <td>0.955075</td>\n",
       "      <td>0.972434</td>\n",
       "      <td>0.936770</td>\n",
       "      <td>0.953851</td>\n",
       "      <td>0.954907</td>\n",
       "      <td>deepSRAMP$_{single}$_maxlen</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUPRC</th>\n",
       "      <td>0.807154</td>\n",
       "      <td>0.736249</td>\n",
       "      <td>0.726165</td>\n",
       "      <td>0.778184</td>\n",
       "      <td>0.789418</td>\n",
       "      <td>0.683276</td>\n",
       "      <td>0.610044</td>\n",
       "      <td>0.753407</td>\n",
       "      <td>deepSRAMP$_{single}$_random</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AUROC</th>\n",
       "      <td>0.967206</td>\n",
       "      <td>0.954290</td>\n",
       "      <td>0.941138</td>\n",
       "      <td>0.955560</td>\n",
       "      <td>0.970154</td>\n",
       "      <td>0.936784</td>\n",
       "      <td>0.950582</td>\n",
       "      <td>0.954189</td>\n",
       "      <td>deepSRAMP$_{single}$_random</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           a549      cd8t      hela     abcam      sysy      molm     ythdf  \\\n",
       "AUPRC  0.418955  0.420442  0.390339  0.278911  0.311095  0.264470  0.238700   \n",
       "AUROC  0.842155  0.838535  0.791095  0.723788  0.785666  0.713264  0.747836   \n",
       "AUPRC  0.425007  0.424914  0.384913  0.272305  0.304974  0.262237  0.237371   \n",
       "AUROC  0.843230  0.839872  0.788713  0.720002  0.787811  0.711557  0.746492   \n",
       "AUPRC  0.793165  0.758831  0.724289  0.765890  0.825184  0.684066  0.649307   \n",
       "AUROC  0.965697  0.957643  0.941825  0.955075  0.972434  0.936770  0.953851   \n",
       "AUPRC  0.807154  0.736249  0.726165  0.778184  0.789418  0.683276  0.610044   \n",
       "AUROC  0.967206  0.954290  0.941138  0.955560  0.970154  0.936784  0.950582   \n",
       "\n",
       "        average                       method  \n",
       "AUPRC  0.347368           DeepPromise_maxlen  \n",
       "AUROC  0.782417           DeepPromise_maxlen  \n",
       "AUPRC  0.345725           DeepPromise_random  \n",
       "AUROC  0.781864           DeepPromise_random  \n",
       "AUPRC  0.758571  deepSRAMP$_{single}$_maxlen  \n",
       "AUROC  0.954907  deepSRAMP$_{single}$_maxlen  \n",
       "AUPRC  0.753407  deepSRAMP$_{single}$_random  \n",
       "AUROC  0.954189  deepSRAMP$_{single}$_random  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b0ff9707-65e5-4682-8597-dc9612833a02",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "08f9af68-b319-40fd-9ef2-75a2d3c9499a",
   "metadata": {},
   "source": [
    "## misc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7944db62-bb9d-4ec0-a1d5-6811ae65ebe8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SRAMP</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a549</th>\n",
       "      <td>0.550586</td>\n",
       "      <td>0.912095</td>\n",
       "      <td>0.686666</td>\n",
       "      <td>0.895816</td>\n",
       "      <td>0.897815</td>\n",
       "      <td>0.658696</td>\n",
       "      <td>0.793165</td>\n",
       "      <td>0.965697</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cd8t</th>\n",
       "      <td>0.527489</td>\n",
       "      <td>0.903842</td>\n",
       "      <td>0.666187</td>\n",
       "      <td>0.881448</td>\n",
       "      <td>0.884308</td>\n",
       "      <td>0.633944</td>\n",
       "      <td>0.758831</td>\n",
       "      <td>0.957643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hela</th>\n",
       "      <td>0.559523</td>\n",
       "      <td>0.821247</td>\n",
       "      <td>0.665580</td>\n",
       "      <td>0.890327</td>\n",
       "      <td>0.880308</td>\n",
       "      <td>0.612132</td>\n",
       "      <td>0.724289</td>\n",
       "      <td>0.941825</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abcam</th>\n",
       "      <td>0.627424</td>\n",
       "      <td>0.719189</td>\n",
       "      <td>0.670180</td>\n",
       "      <td>0.939560</td>\n",
       "      <td>0.912238</td>\n",
       "      <td>0.621757</td>\n",
       "      <td>0.765890</td>\n",
       "      <td>0.955075</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sysy</th>\n",
       "      <td>0.632723</td>\n",
       "      <td>0.870457</td>\n",
       "      <td>0.732791</td>\n",
       "      <td>0.931631</td>\n",
       "      <td>0.924340</td>\n",
       "      <td>0.701916</td>\n",
       "      <td>0.825184</td>\n",
       "      <td>0.972434</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>molm</th>\n",
       "      <td>0.549229</td>\n",
       "      <td>0.714555</td>\n",
       "      <td>0.621079</td>\n",
       "      <td>0.915211</td>\n",
       "      <td>0.889866</td>\n",
       "      <td>0.564548</td>\n",
       "      <td>0.684066</td>\n",
       "      <td>0.936770</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ythdf</th>\n",
       "      <td>0.483259</td>\n",
       "      <td>0.795224</td>\n",
       "      <td>0.601180</td>\n",
       "      <td>0.914968</td>\n",
       "      <td>0.904082</td>\n",
       "      <td>0.572416</td>\n",
       "      <td>0.649307</td>\n",
       "      <td>0.953851</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Precision    Recall        F1  Specificity  Accuracy       MCC  \\\n",
       "SRAMP                                                                   \n",
       "a549    0.550586  0.912095  0.686666     0.895816  0.897815  0.658696   \n",
       "cd8t    0.527489  0.903842  0.666187     0.881448  0.884308  0.633944   \n",
       "hela    0.559523  0.821247  0.665580     0.890327  0.880308  0.612132   \n",
       "abcam   0.627424  0.719189  0.670180     0.939560  0.912238  0.621757   \n",
       "sysy    0.632723  0.870457  0.732791     0.931631  0.924340  0.701916   \n",
       "molm    0.549229  0.714555  0.621079     0.915211  0.889866  0.564548   \n",
       "ythdf   0.483259  0.795224  0.601180     0.914968  0.904082  0.572416   \n",
       "\n",
       "          AUPRC     AUROC  \n",
       "SRAMP                      \n",
       "a549   0.793165  0.965697  \n",
       "cd8t   0.758831  0.957643  \n",
       "hela   0.724289  0.941825  \n",
       "abcam  0.765890  0.955075  \n",
       "sysy   0.825184  0.972434  \n",
       "molm   0.684066  0.936770  \n",
       "ythdf  0.649307  0.953851  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_res = []\n",
    "for i in datasets_1:\n",
    "    test_res += utils.load(f'model/full_400_{i}_single.metrics')\n",
    "\n",
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets_1])\n",
    "t.index.name = 'SRAMP'\n",
    "t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9fc96ccc-a90f-4eb0-be37-e44e7807a666",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SRAMP-rmute</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a549</th>\n",
       "      <td>0.526755</td>\n",
       "      <td>0.936324</td>\n",
       "      <td>0.674212</td>\n",
       "      <td>0.882282</td>\n",
       "      <td>0.888916</td>\n",
       "      <td>0.650399</td>\n",
       "      <td>0.765989</td>\n",
       "      <td>0.963662</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cd8t</th>\n",
       "      <td>0.513484</td>\n",
       "      <td>0.929889</td>\n",
       "      <td>0.661622</td>\n",
       "      <td>0.870991</td>\n",
       "      <td>0.878514</td>\n",
       "      <td>0.633964</td>\n",
       "      <td>0.740773</td>\n",
       "      <td>0.958159</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hela</th>\n",
       "      <td>0.555524</td>\n",
       "      <td>0.838201</td>\n",
       "      <td>0.668196</td>\n",
       "      <td>0.886233</td>\n",
       "      <td>0.879267</td>\n",
       "      <td>0.616991</td>\n",
       "      <td>0.724085</td>\n",
       "      <td>0.942823</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abcam</th>\n",
       "      <td>0.622991</td>\n",
       "      <td>0.695398</td>\n",
       "      <td>0.657206</td>\n",
       "      <td>0.940443</td>\n",
       "      <td>0.910062</td>\n",
       "      <td>0.606840</td>\n",
       "      <td>0.725250</td>\n",
       "      <td>0.950188</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sysy</th>\n",
       "      <td>0.604810</td>\n",
       "      <td>0.884525</td>\n",
       "      <td>0.718400</td>\n",
       "      <td>0.921796</td>\n",
       "      <td>0.917354</td>\n",
       "      <td>0.688644</td>\n",
       "      <td>0.757101</td>\n",
       "      <td>0.966788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>molm</th>\n",
       "      <td>0.538668</td>\n",
       "      <td>0.711489</td>\n",
       "      <td>0.613133</td>\n",
       "      <td>0.911903</td>\n",
       "      <td>0.886588</td>\n",
       "      <td>0.555458</td>\n",
       "      <td>0.645149</td>\n",
       "      <td>0.933411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ythdf</th>\n",
       "      <td>0.457748</td>\n",
       "      <td>0.808763</td>\n",
       "      <td>0.584613</td>\n",
       "      <td>0.904193</td>\n",
       "      <td>0.895518</td>\n",
       "      <td>0.558200</td>\n",
       "      <td>0.603986</td>\n",
       "      <td>0.948703</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Precision    Recall        F1  Specificity  Accuracy       MCC  \\\n",
       "SRAMP-rmute                                                                   \n",
       "a549          0.526755  0.936324  0.674212     0.882282  0.888916  0.650399   \n",
       "cd8t          0.513484  0.929889  0.661622     0.870991  0.878514  0.633964   \n",
       "hela          0.555524  0.838201  0.668196     0.886233  0.879267  0.616991   \n",
       "abcam         0.622991  0.695398  0.657206     0.940443  0.910062  0.606840   \n",
       "sysy          0.604810  0.884525  0.718400     0.921796  0.917354  0.688644   \n",
       "molm          0.538668  0.711489  0.613133     0.911903  0.886588  0.555458   \n",
       "ythdf         0.457748  0.808763  0.584613     0.904193  0.895518  0.558200   \n",
       "\n",
       "                AUPRC     AUROC  \n",
       "SRAMP-rmute                      \n",
       "a549         0.765989  0.963662  \n",
       "cd8t         0.740773  0.958159  \n",
       "hela         0.724085  0.942823  \n",
       "abcam        0.725250  0.950188  \n",
       "sysy         0.757101  0.966788  \n",
       "molm         0.645149  0.933411  \n",
       "ythdf        0.603986  0.948703  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_res = []\n",
    "for i in datasets_1:\n",
    "    test_res += utils.load(f'model/full_400_{i}_single_rmute.metrics')\n",
    "\n",
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets_1])\n",
    "t.index.name = 'SRAMP-rmute'\n",
    "t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "30a5ef60-c1db-4702-abd5-dabcb2307f97",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SRAMP-nom6a</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a549</th>\n",
       "      <td>0.342575</td>\n",
       "      <td>0.890669</td>\n",
       "      <td>0.494827</td>\n",
       "      <td>0.760809</td>\n",
       "      <td>0.776751</td>\n",
       "      <td>0.458626</td>\n",
       "      <td>0.534733</td>\n",
       "      <td>0.898815</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cd8t</th>\n",
       "      <td>0.353193</td>\n",
       "      <td>0.896896</td>\n",
       "      <td>0.506807</td>\n",
       "      <td>0.759495</td>\n",
       "      <td>0.777045</td>\n",
       "      <td>0.468016</td>\n",
       "      <td>0.540440</td>\n",
       "      <td>0.901816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hela</th>\n",
       "      <td>0.434160</td>\n",
       "      <td>0.737471</td>\n",
       "      <td>0.546555</td>\n",
       "      <td>0.836954</td>\n",
       "      <td>0.822526</td>\n",
       "      <td>0.469437</td>\n",
       "      <td>0.533366</td>\n",
       "      <td>0.865077</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abcam</th>\n",
       "      <td>0.299256</td>\n",
       "      <td>0.533151</td>\n",
       "      <td>0.383343</td>\n",
       "      <td>0.823315</td>\n",
       "      <td>0.787341</td>\n",
       "      <td>0.283185</td>\n",
       "      <td>0.316684</td>\n",
       "      <td>0.758892</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sysy</th>\n",
       "      <td>0.332535</td>\n",
       "      <td>0.813599</td>\n",
       "      <td>0.472109</td>\n",
       "      <td>0.779029</td>\n",
       "      <td>0.783149</td>\n",
       "      <td>0.422474</td>\n",
       "      <td>0.416343</td>\n",
       "      <td>0.864815</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>molm</th>\n",
       "      <td>0.312403</td>\n",
       "      <td>0.616183</td>\n",
       "      <td>0.414603</td>\n",
       "      <td>0.803922</td>\n",
       "      <td>0.780208</td>\n",
       "      <td>0.322671</td>\n",
       "      <td>0.343550</td>\n",
       "      <td>0.778574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ythdf</th>\n",
       "      <td>0.215208</td>\n",
       "      <td>0.746145</td>\n",
       "      <td>0.334063</td>\n",
       "      <td>0.727905</td>\n",
       "      <td>0.729563</td>\n",
       "      <td>0.293333</td>\n",
       "      <td>0.264559</td>\n",
       "      <td>0.802142</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Precision    Recall        F1  Specificity  Accuracy       MCC  \\\n",
       "SRAMP-nom6a                                                                   \n",
       "a549          0.342575  0.890669  0.494827     0.760809  0.776751  0.458626   \n",
       "cd8t          0.353193  0.896896  0.506807     0.759495  0.777045  0.468016   \n",
       "hela          0.434160  0.737471  0.546555     0.836954  0.822526  0.469437   \n",
       "abcam         0.299256  0.533151  0.383343     0.823315  0.787341  0.283185   \n",
       "sysy          0.332535  0.813599  0.472109     0.779029  0.783149  0.422474   \n",
       "molm          0.312403  0.616183  0.414603     0.803922  0.780208  0.322671   \n",
       "ythdf         0.215208  0.746145  0.334063     0.727905  0.729563  0.293333   \n",
       "\n",
       "                AUPRC     AUROC  \n",
       "SRAMP-nom6a                      \n",
       "a549         0.534733  0.898815  \n",
       "cd8t         0.540440  0.901816  \n",
       "hela         0.533366  0.865077  \n",
       "abcam        0.316684  0.758892  \n",
       "sysy         0.416343  0.864815  \n",
       "molm         0.343550  0.778574  \n",
       "ythdf        0.264559  0.802142  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_res = []\n",
    "for i in datasets_1:\n",
    "    test_res += utils.load(f'model/full_400_{i}_single_.metrics')\n",
    "\n",
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets_1])\n",
    "t.index.name = 'SRAMP-nom6a'\n",
    "t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "0444b7a1-5b29-4576-86e8-6acd3c3f50dc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>DP</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a549</th>\n",
       "      <td>0.289983</td>\n",
       "      <td>0.804565</td>\n",
       "      <td>0.426313</td>\n",
       "      <td>0.724325</td>\n",
       "      <td>0.734176</td>\n",
       "      <td>0.366232</td>\n",
       "      <td>0.418955</td>\n",
       "      <td>0.842155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cd8t</th>\n",
       "      <td>0.293629</td>\n",
       "      <td>0.798350</td>\n",
       "      <td>0.429347</td>\n",
       "      <td>0.718781</td>\n",
       "      <td>0.728944</td>\n",
       "      <td>0.362546</td>\n",
       "      <td>0.420442</td>\n",
       "      <td>0.838535</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hela</th>\n",
       "      <td>0.313542</td>\n",
       "      <td>0.674050</td>\n",
       "      <td>0.427996</td>\n",
       "      <td>0.749659</td>\n",
       "      <td>0.738693</td>\n",
       "      <td>0.322095</td>\n",
       "      <td>0.390339</td>\n",
       "      <td>0.791095</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abcam</th>\n",
       "      <td>0.234441</td>\n",
       "      <td>0.580343</td>\n",
       "      <td>0.333969</td>\n",
       "      <td>0.731799</td>\n",
       "      <td>0.713022</td>\n",
       "      <td>0.223042</td>\n",
       "      <td>0.278911</td>\n",
       "      <td>0.723788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sysy</th>\n",
       "      <td>0.253255</td>\n",
       "      <td>0.684056</td>\n",
       "      <td>0.369655</td>\n",
       "      <td>0.727078</td>\n",
       "      <td>0.721951</td>\n",
       "      <td>0.285115</td>\n",
       "      <td>0.311095</td>\n",
       "      <td>0.785666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>molm</th>\n",
       "      <td>0.232547</td>\n",
       "      <td>0.574428</td>\n",
       "      <td>0.331067</td>\n",
       "      <td>0.725921</td>\n",
       "      <td>0.706785</td>\n",
       "      <td>0.215354</td>\n",
       "      <td>0.264470</td>\n",
       "      <td>0.713264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ythdf</th>\n",
       "      <td>0.173946</td>\n",
       "      <td>0.696502</td>\n",
       "      <td>0.278371</td>\n",
       "      <td>0.669237</td>\n",
       "      <td>0.671715</td>\n",
       "      <td>0.218523</td>\n",
       "      <td>0.238700</td>\n",
       "      <td>0.747836</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Precision    Recall        F1  Specificity  Accuracy       MCC  \\\n",
       "DP                                                                      \n",
       "a549    0.289983  0.804565  0.426313     0.724325  0.734176  0.366232   \n",
       "cd8t    0.293629  0.798350  0.429347     0.718781  0.728944  0.362546   \n",
       "hela    0.313542  0.674050  0.427996     0.749659  0.738693  0.322095   \n",
       "abcam   0.234441  0.580343  0.333969     0.731799  0.713022  0.223042   \n",
       "sysy    0.253255  0.684056  0.369655     0.727078  0.721951  0.285115   \n",
       "molm    0.232547  0.574428  0.331067     0.725921  0.706785  0.215354   \n",
       "ythdf   0.173946  0.696502  0.278371     0.669237  0.671715  0.218523   \n",
       "\n",
       "          AUPRC     AUROC  \n",
       "DP                         \n",
       "a549   0.418955  0.842155  \n",
       "cd8t   0.420442  0.838535  \n",
       "hela   0.390339  0.791095  \n",
       "abcam  0.278911  0.723788  \n",
       "sysy   0.311095  0.785666  \n",
       "molm   0.264470  0.713264  \n",
       "ythdf  0.238700  0.747836  "
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_res = []\n",
    "for i in datasets_1:\n",
    "    test_res += [utils.load(f'model/full_400_{i}_dp.metrics')[-1]]\n",
    "\n",
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets_1])\n",
    "t.index.name = 'DP'\n",
    "t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e94e4397-827a-4994-b48b-da3a583c4491",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "74c607a6-5bdc-43d0-8b4a-a430c64d8ec4",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = utils.load('data/GSE/compare_mature.data')\n",
    "ythdf = utils.load('data/GSE/compare_mature_ythdf_test.data')\n",
    "datasets = ['abcam', 'sysy', 'cd8t', 'a549', 'molm', 'hela', 'ythdf']\n",
    "datasets_1 = ['a549', 'cd8t', 'hela', 'abcam', 'sysy', 'molm', 'ythdf']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "e49d30cd-bd4d-4e63-ac34-9a9291221064",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating test data - [abcam]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:01<00:00, 7232.77it/s]\n",
      "100%|██████████| 20681/20681 [00:08<00:00, 2473.39it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model - [abcam]\n",
      "Start to train - [abcam]\n",
      "[Test] ACC: 78.9%, Loss: 0.481191, AUROC: 0.7579, AUPRC: 0.3226, Size: 20681, Ratio: 0.12398\n",
      "Generating test data - [sysy]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:00<00:00, 10848.36it/s]\n",
      "100%|██████████| 14314/14314 [00:05<00:00, 2504.82it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model - [sysy]\n",
      "Start to train - [sysy]\n",
      "[Test] ACC: 77.9%, Loss: 0.484867, AUROC: 0.8643, AUPRC: 0.4186, Size: 14314, Ratio: 0.11918\n",
      "Generating test data - [cd8t]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:02<00:00, 4177.07it/s]\n",
      "100%|██████████| 36070/36070 [00:16<00:00, 2142.74it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model - [cd8t]\n",
      "Start to train - [cd8t]\n",
      "[Test] ACC: 77.4%, Loss: 0.510465, AUROC: 0.9025, AUPRC: 0.5449, Size: 36070, Ratio: 0.12772\n",
      "Generating test data - [a549]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:02<00:00, 4749.24it/s]\n",
      "100%|██████████| 40681/40681 [00:18<00:00, 2250.66it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model - [a549]\n",
      "Start to train - [a549]\n",
      "[Test] ACC: 77.2%, Loss: 0.506325, AUROC: 0.8988, AUPRC: 0.5352, Size: 40681, Ratio: 0.12276\n",
      "Generating test data - [molm]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:01<00:00, 5612.25it/s]\n",
      "100%|██████████| 33559/33559 [00:13<00:00, 2476.12it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model - [molm]\n",
      "Start to train - [molm]\n",
      "[Test] ACC: 77.9%, Loss: 0.495369, AUROC: 0.7769, AUPRC: 0.3428, Size: 33559, Ratio: 0.12631\n",
      "Generating test data - [hela]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10331/10331 [00:05<00:00, 1873.10it/s]\n",
      "100%|██████████| 111436/111436 [00:54<00:00, 2048.85it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model - [hela]\n",
      "Start to train - [hela]\n",
      "[Test] ACC: 81.6%, Loss: 0.426650, AUROC: 0.8633, AUPRC: 0.5316, Size: 111436, Ratio: 0.14503\n"
     ]
    }
   ],
   "source": [
    "test_res = []\n",
    "for name in datasets_1[:-1]:\n",
    "    i = name\n",
    "    print(f'Generating test data - [{name}]')\n",
    "    if name == 'ythdf':\n",
    "        testds = df2ds(ythdf, downsample=False, pos_label=f'{i}_test', neg_label=f'{i}_test_neg')\n",
    "    else:\n",
    "        testds = df2ds(df, downsample=False, pos_label=f'{i}_test', neg_label=f'{i}_test_neg')\n",
    "    testdl = ds2dl(testds, drop_last=False, num_workers=0)\n",
    "\n",
    "    # print(f'Generating train data - [{name}]')\n",
    "    # trainds = df2ds(df, downsample=False, pos_label=f'{i}_train', neg_label=f'{i}_train_neg')\n",
    "    # traindl = ds2dl(trainds, drop_last=True, num_workers=8)\n",
    "    \n",
    "    print(f'Building model - [{name}]')\n",
    "    device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "    model = SRAMP().to(device)\n",
    "    model.load_state_dict(torch.load(f'model/full_400_{i}_single_rmute.model', weights_only=True))\n",
    "    loss_fn = TriLoss()\n",
    "    optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)\n",
    "    scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer, 1, 2)\n",
    "\n",
    "    print(f'Start to train - [{name}]')\n",
    "    # train(model, loss_fn, optimizer, scheduler, traindl, testdl, epochs=1, device=device)\n",
    "    test_res += [test_loop(testdl, model, device, loss_fn)]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "823d417d-e3c7-4e70-b213-cda38f582f41",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>MCC</th>\n",
       "      <th>AUPRC</th>\n",
       "      <th>AUROC</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>abcam</th>\n",
       "      <td>0.302448</td>\n",
       "      <td>0.539782</td>\n",
       "      <td>0.387675</td>\n",
       "      <td>0.823812</td>\n",
       "      <td>0.788598</td>\n",
       "      <td>0.288665</td>\n",
       "      <td>0.322563</td>\n",
       "      <td>0.757899</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sysy</th>\n",
       "      <td>0.327440</td>\n",
       "      <td>0.814185</td>\n",
       "      <td>0.467048</td>\n",
       "      <td>0.773715</td>\n",
       "      <td>0.778538</td>\n",
       "      <td>0.417132</td>\n",
       "      <td>0.418565</td>\n",
       "      <td>0.864255</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cd8t</th>\n",
       "      <td>0.350250</td>\n",
       "      <td>0.897330</td>\n",
       "      <td>0.503839</td>\n",
       "      <td>0.756253</td>\n",
       "      <td>0.774272</td>\n",
       "      <td>0.464950</td>\n",
       "      <td>0.544880</td>\n",
       "      <td>0.902489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>a549</th>\n",
       "      <td>0.338332</td>\n",
       "      <td>0.894473</td>\n",
       "      <td>0.490960</td>\n",
       "      <td>0.755205</td>\n",
       "      <td>0.772302</td>\n",
       "      <td>0.455354</td>\n",
       "      <td>0.535243</td>\n",
       "      <td>0.898826</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>molm</th>\n",
       "      <td>0.308697</td>\n",
       "      <td>0.607926</td>\n",
       "      <td>0.409470</td>\n",
       "      <td>0.803172</td>\n",
       "      <td>0.778509</td>\n",
       "      <td>0.315917</td>\n",
       "      <td>0.342821</td>\n",
       "      <td>0.776856</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hela</th>\n",
       "      <td>0.423092</td>\n",
       "      <td>0.743905</td>\n",
       "      <td>0.539402</td>\n",
       "      <td>0.827928</td>\n",
       "      <td>0.815742</td>\n",
       "      <td>0.461983</td>\n",
       "      <td>0.531621</td>\n",
       "      <td>0.863252</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Precision    Recall        F1  Specificity  Accuracy       MCC  \\\n",
       "                                                                        \n",
       "abcam   0.302448  0.539782  0.387675     0.823812  0.788598  0.288665   \n",
       "sysy    0.327440  0.814185  0.467048     0.773715  0.778538  0.417132   \n",
       "cd8t    0.350250  0.897330  0.503839     0.756253  0.774272  0.464950   \n",
       "a549    0.338332  0.894473  0.490960     0.755205  0.772302  0.455354   \n",
       "molm    0.308697  0.607926  0.409470     0.803172  0.778509  0.315917   \n",
       "hela    0.423092  0.743905  0.539402     0.827928  0.815742  0.461983   \n",
       "\n",
       "          AUPRC     AUROC  \n",
       "                           \n",
       "abcam  0.322563  0.757899  \n",
       "sysy   0.418565  0.864255  \n",
       "cd8t   0.544880  0.902489  \n",
       "a549   0.535243  0.898826  \n",
       "molm   0.342821  0.776856  \n",
       "hela   0.531621  0.863252  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = np.array([[i[-1]['1.0']['precision'], i[-1]['1.0']['recall'], i[-1]['1.0']['f1-score'], i[-1]['0.0']['recall'], i[-1]['accuracy'], i[-2], i[2], i[1]] for i in (test_res)])#.shape#.reshape((7, 5, -1))\n",
    "t = pd.DataFrame(data, columns=['Precision', 'Recall', 'F1', 'Specificity', 'Accuracy', 'MCC', 'AUPRC', 'AUROC'], index=[i for i in datasets[:-1]])\n",
    "t.index.name = ''\n",
    "t"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:sramp]",
   "language": "python",
   "name": "conda-env-sramp-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}