From 6adfa514838b710dc50a4ce24022418f64fb2b2d Mon Sep 17 00:00:00 2001
From: RAHenriksen <rah@hotmail.dk>
Date: Thu, 2 Jan 2025 11:34:46 +0100
Subject: [PATCH] udpate

---
 bifrost_chewbbaca/pipeline_blat.smk        |  2 ++
 bifrost_chewbbaca/rule__mmseqs_genecall.py | 20 ++++++++++++--------
 2 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/bifrost_chewbbaca/pipeline_blat.smk b/bifrost_chewbbaca/pipeline_blat.smk
index 44c995c..b8026d6 100755
--- a/bifrost_chewbbaca/pipeline_blat.smk
+++ b/bifrost_chewbbaca/pipeline_blat.smk
@@ -119,6 +119,8 @@ rule blat_gene_call:
         gene_call_results = directory(f"{component['name']}/blat_gene_call_results"),
         gene_calls = f"{component['name']}/blat_gene_call_results/gene_calls.fa",
         gene_call_done = f"{component['name']}/blat_gene_call_done"
+    message:
+        f"Running step:{rule_name}, outputs: {component['name']}/blat_gene_call_results"
     script:
         os.path.join(os.path.dirname(workflow.snakefile), "rule__blat_genecall.py")
 
diff --git a/bifrost_chewbbaca/rule__mmseqs_genecall.py b/bifrost_chewbbaca/rule__mmseqs_genecall.py
index df6b534..78c1a9a 100755
--- a/bifrost_chewbbaca/rule__mmseqs_genecall.py
+++ b/bifrost_chewbbaca/rule__mmseqs_genecall.py
@@ -78,23 +78,27 @@ def parse_mmseqs_output(output_tsv, assembly_sequences):
 
     return alleles
 
-def run_mmseqs_and_parse(query_fa, db_fa, output_tsv, assembly_sequences):
+def run_mmseqs_and_parse(query_fa, db, output_tsv, assembly_sequences):
     """
     Runs mmseqs and processes the outputs.
     """
     # Define the mmseqs command
     mmseqs_cmd = [
-        "mmseqs easy-search",
-        db_fa,
+        "mmseqs",
+        "easy-linsearch",
         query_fa,
+        db,
         output_tsv,
         "tmp",
-        "--search-type",'4',
-        "--format-mode",'0'
-        "-format-output",'query,target,tlen,pident,alnlen,mismatch,gapopen,qstart,qend,tstart,tend,evalue,bits'
-        "--min-seq-id",0.9
+        "--search-type",'3',
+        "--format-mode",'0',
+        "--format-output",'query,target,tlen,pident,alnlen,mismatch,gapopen,qstart,qend,tstart,tend,evalue,bits',
+        "--min-seq-id",'0.9',
+        "--threads",'6',
+        "--remove-tmp-files",'1',
     ]
-    # mmseqs easy-search assemblatron__v2.3.3/test_cdiff_single___2405W4378.fasta all_Clostridioides_loci.fa alnRes.tsv tmp --search-type 4 --format-mode 0 --format-output "query,target,tlen,pident,alnlen,mismatch,gapopen,qstart,qend,tstart,tend,evalue,bits" --min-seq-id 0.9
+    print(f"msmeqs commands {mmseqs_cmd}")
+    #mmseqs easy-linsearch assemblatron__v2.3.3/test_cdiff_single___2405W4378.fasta all_Clostridioides_loci_db alnRes.tsv tmp --search-type 3 --threads 6 --min-seq-id 0.9 --remove-tmp-files 1 --format-mode 0 --format-output "query,target,qlen,pident,alnlen,mismatch,gapopen,qstart,qend,tstart,tend,evalue,bits"
 
     # Run mmseqs
     try: