analysis: mmseqs2: bug fix (wrong db type) (issue #119); metap - whether faa output is tmp

config/gdb/config.yaml

@@ -34,8 +34,8 @@ data:
             r1: "/mnt/isilon/projects/ecosystem_biology/ONT_pilot/GDB_2019/metat/sr/FastSelectFull1_MT_Rashi_S14_R1_001.fastq.gz" # leave empty if no data, i.e. ""
             r2: "/mnt/isilon/projects/ecosystem_biology/ONT_pilot/GDB_2019/metat/sr/FastSelectFull1_MT_Rashi_S14_R2_001.fastq.gz" # leave empty if no data, i.e. ""
     # Meta-proteomics
-    # metap:
-        # TODO
+    metap:
+        "" # TODO
 
 ############################################################
 # TOOLS

workflow/rules/analysis.smk

@@ -339,7 +339,7 @@ rule analysis_mmseqs2:
         )
     output:
         tsv=os.path.join(RESULTS_DIR, "analysis/mmseqs2/clusters.tsv"),
-        faa=temp(os.path.join(RESULTS_DIR, "analysis/mmseqs2/proteins.faa")),
+        faa=temp(os.path.join(RESULTS_DIR, "analysis/mmseqs2/proteins.faa")) if "metap" not in config["data"] else os.path.join(RESULTS_DIR, "analysis/mmseqs2/proteins.faa"),
         tmpdir=temp(directory(os.path.join(RESULTS_DIR, "analysis/mmseqs2/tmp")))
     log:
         "logs/analysis.mmseqs2.log"
@@ -358,7 +358,7 @@ rule analysis_mmseqs2:
         # put all proteins together
         "for f in {input}; do sed 's/^>/>'$(basename $(dirname ${{f}}))'__/' ${{f}}; done > {output.faa} && "
         # DB, cluster, TSV output
-        "mmseqs createdb --dbtype 2 -v 3 {output.faa} {params.db} && "
+        "mmseqs createdb --dbtype 1 -v 3 {output.faa} {params.db} && "
         "mmseqs linclust --threads {threads} {params.db} {params.cl} {output.tmpdir} && "
         "mmseqs createtsv {params.db} {params.db} {params.cl} {output.tsv} && "
         "date) &> {log}"