From 503363aa1be9bf89a03e10f020c5e273d6bf4852 Mon Sep 17 00:00:00 2001
From: Rongting <rthuang@foxmail.com>
Date: Sat, 15 Jun 2024 15:22:10 +0800
Subject: [PATCH] enviroment and installation

---
 docs/FAQ.rst                   | 12 +++++++++---
 docs/release.rst               |  8 ++++++++
 setup.py                       |  2 +-
 xclone/model/_RDR_smoothing.py | 35 ++++++++++++++++++++++++++++++----
 xclone/version.py              |  2 +-
 5 files changed, 50 insertions(+), 9 deletions(-)

diff --git a/docs/FAQ.rst b/docs/FAQ.rst
index b39bc65..a2698d3 100644
--- a/docs/FAQ.rst
+++ b/docs/FAQ.rst
@@ -11,7 +11,7 @@ Top Questions
 Input and Output
 ----------------
 
-CNV Modules
+CNA Modules
 -----------
 
 Parameters
@@ -19,6 +19,8 @@ Parameters
 
 Version 0.3.4 specific issues
 -----------------------------
+Not support python >3.7.
+Recommend XClone version 0.3.6 for Python >3.7 environment.
 
 FileNotFoundError
 ~~~~~~~~~~~~~~~~~
@@ -74,7 +76,9 @@ You may download the `anno_data` from the following URL and place the files unde
 Version 0.3.5 specific issues
 -----------------------------
 
-Not reproted so far.
+Not support python >3.7.
+Recommend XClone version 0.3.6 for Python >3.7 environment.
+No other issues reproted so far.
 
 
 Dependency
@@ -89,6 +93,8 @@ You may encounter an error indicating that the `requests` module is not found (i
 
     pip install requests
 
+This Dependency issues solved in XClone version 0.3.6.
+
 ModuleNotFoundError: No module named 'importlib.metadata'
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
@@ -178,4 +184,4 @@ The most import step you may try is:
 
     pip install scanpy
 
-
+This Dependency issues solved in XClone version 0.3.6.
diff --git a/docs/release.rst b/docs/release.rst
index 96eabae..3820baf 100644
--- a/docs/release.rst
+++ b/docs/release.rst
@@ -1,6 +1,14 @@
 Release History
 ===============
 
+Version 0.3.6
+-------------
+- update Dependency for installation
+- [Python version supporting] XClone support Python >3.7
+  adding a nonzero scalar to a sparse array is not supported by `scipy` in Python > 3.7 (which exists in RDR_smothing functions).
+
+
+
 Version 0.3.5
 -------------
 - [BAF] add allele flip status for each gene in Local phasing and Global phasing
diff --git a/setup.py b/setup.py
index d74cd9e..b45f8cf 100644
--- a/setup.py
+++ b/setup.py
@@ -19,7 +19,7 @@
     long_description = f.read()
 
 reqs = ['numpy>=1.18', 'scipy', 'matplotlib', 'anndata',
-        'statsmodels', 'scanpy', 'h5py', 'palettable']
+        'statsmodels', 'scanpy', 'h5py', 'palettable', 'requests', 'importlib-metadata']
 
 # seaborn
 
diff --git a/xclone/model/_RDR_smoothing.py b/xclone/model/_RDR_smoothing.py
index d3afe92..a992c5e 100644
--- a/xclone/model/_RDR_smoothing.py
+++ b/xclone/model/_RDR_smoothing.py
@@ -5,6 +5,8 @@
 # update: 2022/07/26
 
 import numpy as np
+import scanpy as sc
+import scipy.sparse as sp
 from .smoothing import WMA_smooth, KNN_smooth
 
 ## Smoothing strategy
@@ -25,12 +27,37 @@ def RDR_smoothing_base(Xdata,
     """
     ## preprocess
     ## normalization [follow scanpy pipeline]
+    # Xdata_norm.X = np.log(Xdata_norm.X/Xdata_norm.X.sum(1) * 10000 + 1) 
+    # Notes: this only supported by python<3.7 if issparse
+    # sc.pp.normalize_total from the Scanpy library supports both dense and sparse matrices. 
+    
     Xdata_norm = Xdata.copy()
-    Xdata_norm.X = np.log(Xdata_norm.X/Xdata_norm.X.sum(1) * 10000 + 1)
-
     _is_ref = Xdata_norm.obs[cell_anno_key] == ref_celltype
-    Xdata_norm.var['ref_mean'] = Xdata_norm[_is_ref, :].X.mean(0)
-    Xdata_norm.var['ref_var'] = Xdata_norm[_is_ref, :].X.var(0)
+
+    # Normalize and log transform using scanpy functions
+    sc.pp.normalize_total(Xdata_norm, target_sum=10000)
+    sc.pp.log1p(Xdata_norm)
+
+    # Check if Xdata_norm.X is a sparse matrix
+    if sp.issparse(Xdata_norm.X):
+        # # Perform operations in a way that supports sparse matrices
+        # sums = np.array(Xdata_norm.X.sum(axis=1)).flatten()
+        # Xdata_norm.X = Xdata_norm.X.multiply(1 / sums[:, None])
+        # Xdata_norm.X = Xdata_norm.X.multiply(10000)
+        # # Create a sparse matrix of ones with the same shape
+        # ones_sparse = sp.csr_matrix(np.ones(Xdata_norm.X.shape))
+        # # Add the sparse matrix of ones to the original sparse matrix
+        # Xdata_norm.X = Xdata_norm.X + ones_sparse
+        # Xdata_norm.X.data = np.log(Xdata_norm.X.data)
+        
+        Xdata_norm.var['ref_mean'] = np.array(Xdata_norm[_is_ref, :].X.mean(axis=0)).flatten()
+        ref_data_dense = Xdata_norm[_is_ref, :].X.toarray()
+        Xdata_norm.var['ref_var'] = ref_data_dense.var(axis=0)
+    else:
+        # Xdata_norm.X = np.log(Xdata_norm.X / Xdata_norm.X.sum(axis=1) * 10000 + 1)
+        Xdata_norm.var['ref_mean'] = Xdata_norm[_is_ref, :].X.mean(0)
+        Xdata_norm.var['ref_var'] = Xdata_norm[_is_ref, :].X.var(0)
+      
 
     adata_tmp = Xdata_norm.copy()
     adata_tmp.X = adata_tmp.X - Xdata_norm.var['ref_mean'].values.reshape(1, -1)
diff --git a/xclone/version.py b/xclone/version.py
index a8d4557..d7b30e1 100644
--- a/xclone/version.py
+++ b/xclone/version.py
@@ -1 +1 @@
-__version__ = "0.3.5"
+__version__ = "0.3.6"