metabat

catalogue.smk

@@ -54,7 +54,7 @@ rule cd_hit:
 		"-i {input} "
 		"-o {output.ffn} "
 		"-c 0.95 "
-		"-G 1 "
+		"-g 1 "
 		"-aS 0.90 "
 		"-d 0 "
 		"-M 0 "
@@ -106,3 +106,56 @@ rule faa_catalogue:
 		"{input.faa} "
 		" ; "
 		"conda deactivate"
+
+checkpoint metabat:
+	input:
+		unpack(contigs_input)
+	output:
+		clusters = directory("work/metabat/{sample}")
+	params:
+		#output = "{output}/{sample}",
+		output = "work/metabat/{sample}/{sample}",
+		clustrsize = 200000
+	threads:
+		8
+	shell:
+		"conda activate metabat2-2.12.1 "
+		" ; "
+		"metabat2 "
+		"--inFile {input} "
+		"--outFile {params.output} "
+		"--minContig 5000 "
+		"--minClsSize {params.clustrsize} "
+		"--numThreads {threads} "
+		"--onlyLabel "
+		"--unbinned "
+		"--verbose "
+		" ; "
+		"conda deactivate "
+
+rule metabat_kaiju:
+	input:
+		clustr = "work/metabat/{sample}/{sample}.{num}",
+		kaiju = "report/contigs_{sample}-taxNames.tsv"
+	output:
+		"work/metabat/{sample}/{sample}.{num}_kaiju.tsv"
+	shell:
+		"grep "
+		"--no-messages "
+		"--word-regexp "
+		"--file {input.clustr} "
+		"{input.kaiju} "
+		" > "
+		"{output} "
+
+def aggregate_metabat(wildcards):
+	checkpoint_output = checkpoints.metabat.get(**wildcards).output.clusters
+	return expand("work/metabat/{sample}/{sample}.{i}_kaiju.tsv", sample=wildcards.sample, i=glob_wildcards(os.path.join(checkpoint_output, "{sample}.{i}")).i)
+
+rule resume_metabat:
+	input:
+		aggregate_metabat
+	output:
+		done = temp("work/metabat/{sample}.done")
+	shell:
+		"touch {output.done} "

cluster.json

@@ -30,5 +30,10 @@
 	{
 		"queue" : "highmem.q,maiage.q",
 		"cluster" : "-l h_vmem=4G"
+	},
+	"metabat_kaiju" :
+	{
+		"queue" : "highmem.q,maiage.q",
+		"cluster" : "-l big_mem"
 	}
 }

global.smk

@@ -16,6 +16,7 @@ rule all:
 		"report/diamond_swissprot_coassembly.tsv",
 		"report/contigs_coassembly-taxNames.tsv",
 		"report/genes_coassembly-taxNames.tsv",
+		"work/metabat/coassembly.done",
 		#catalogue
 		"report/quast_results/report.html",
 		expand("report/count-contigs-catalogue-{sample}.tsv", sample=config["SAMPLES"]),