vhm

global.smk

@@ -28,6 +28,7 @@ rule all:
 		#expand("work/addon/catalogue.{sample}.unmapped.taxNames.tsv", sample=config["SAMPLES"]),
 		#virome
 		expand("report/viromeQC-{sample}.txt", sample=config["SAMPLES"]),
+		expand("work/virhostmatcher/{sample}/done", sample=config["SAMPLES"]),
 
 
 include: "../workflow_metagenomics/quality.smk"

virome.smk

@@ -19,3 +19,29 @@ rule viromeQC:
 		"--tempdir /projet/tmp/ "
 		"&& "
 		"conda deactivate "
+
+rule virhostmatcher:
+	input:
+		unpack(contigs_input)
+	output:
+		done = "work/virhostmatcher/{sample}/done"
+	params:
+		splittedGenomes = "work/virhostmatcher/{sample}/splittedGenomes",
+		output = lambda wildcards, output: os.path.dirname(str(output))
+	shell:
+		"mkdir -p {params.splittedGenomes} "
+		"&& "
+		"awk '/^>/ {{if(x>0) {{close(outname); x=0}} match($0, \">([^| ]*)\", record);outname=sprintf(\"{params.splittedGenomes}/%s.fa\",record[1]); if (x>0) {{print >> outname}} else {{print > outname;}} x++; next;}} {{if(x>0) print >> outname;}}' {input} " # https://github.com/soedinglab/WIsH#tricks
+		"&& "
+		# "conda activate vhm "
+		# "&& "
+		"vhm.py "
+		"--virusFaDir {params.splittedGenomes} "
+		"--hostFaDir /projet/irstea/WIsH_2020/Host/ "
+		"--out {params.output} "
+		"--taxa /projet/irstea/WIsH_2020/Host.txt "
+		"--d2star 0 "
+		# "&& "
+		# "conda deactivate "
+		"&& "
+		"touch {output.done} "