cdhit: changed renaming rule (issue #75), use '-s2 0.5' (issue #68)

workflow/rules/analysis.smk

@@ -28,27 +28,29 @@ rule analysis_quast:
 # Proteins
 
 # for CD-HIT
-rule analysis_bbmap_rename:
+rule analysis_cdhit_rename:
     input:
         os.path.join(RESULTS_DIR, "annotation/prodigal/{rtype}/{tool}/proteins.faa")
     output:
         temp(os.path.join(RESULTS_DIR, "analysis/cdhit/{rtype}_{tool}.faa"))
-    log:
-        "logs/analysis_bbmap_rename_{rtype}.{tool}.log"
+    # log:
+    #     "logs/analysis_bbmap_rename_{rtype}.{tool}.log"
     wildcard_constraints:
         rtype="|".join(READ_TYPES),
         tool="|".join(ASSEMBLERS)
-    threads:
-        config["bbmap"]["threads"]
-    params:
-        # TODO: is this prefix okay?
-        prefix=lambda wildcards: "%s_%s" % (wildcards.rtype, wildcards.tool)
-    conda:
-        os.path.join(ENV_DIR, "bbmap.yaml")
+    threads: 1
+        # config["bbmap"]["threads"]
+    # params:
+    #     # TODO: is this prefix okay?
+    #     prefix=lambda wildcards: "%s_%s" % (wildcards.rtype, wildcards.tool)
+    # conda:
+    #     os.path.join(ENV_DIR, "bbmap.yaml")
     message:
-        "BBMAP: rename FASTA entries in {input}"
+        # "BBMAP: rename FASTA entries in {input}"
+        "CDHIT: rename FASTA entries in {input}"
     shell:
-        "(date && rename.sh in={input} out={output} prefix={params.prefix} ignorejunk=t && date) &> {log}"
+        # "(date && rename.sh in={input} out={output} prefix={params.prefix} ignorejunk=t && date) &> {log}"
+        "sed 's/^>/>{wildcards.rtype}__{wildcards.tool}__/' {input} > {output}"
 
 # https://github.com/weizhongli/cdhit/wiki/3.-User's-Guide#CDHIT2D
 rule analysis_cdhit:
@@ -73,8 +75,8 @@ rule analysis_cdhit:
         "CD-HIT: {input}"
     shell:
         "(date && "
-        "cd-hit-2d -i {input.faa1} -i2 {input.faa2} -o {output.faa12} -c 0.9 -n 5 -d 0 -M 16000 -s2 0.9 -T {threads} && "
-        "cd-hit-2d -i {input.faa2} -i2 {input.faa1} -o {output.faa21} -c 0.9 -n 5 -d 0 -M 16000 -s2 0.9 -T {threads} && "
+        "cd-hit-2d -i {input.faa1} -i2 {input.faa2} -o {output.faa12} -c 0.9 -n 5 -d 0 -M 16000 -s2 0.5 -T {threads} && "
+        "cd-hit-2d -i {input.faa2} -i2 {input.faa1} -o {output.faa21} -c 0.9 -n 5 -d 0 -M 16000 -s2 0.5 -T {threads} && "
         "date) &> {log}"
 
 ##################################################