khmer

assembly.smk

+def khmer_input(wildcards):
+	if config["SORTMERNA"]:
+		return {"R1R2": "work/sortmerna/{wildcards.reads}_mRNA.fastq.gz".format(wildcards=wildcards)}
+	else:
+		return {"R1R2": "DATA/trim/{wildcards.reads}_R1R2.fastq.gz".format(wildcards=wildcards)}
+
+rule khmer:
+	input:
+		unpack(khmer_input)
+	output:
+		R1R2 = "work/khmer/{reads}_R1R2.fastq.gz",
+		log = "work/khmer/{reads}.log"
+	threads:
+		10
+	params:
+		mem_tot = 90 * 1e9,
+		K = 32,
+		C = 100
+	shell:
+		"normalize-by-median.py "
+		"--ksize {params.K} "
+		"--max-memory-usage {params.mem_tot} "
+		"--cutoff {params.C} "
+		"--paired "
+		"--report {output.log} "
+		"--output {output.R1R2} "
+		"--gzip "
+		"{input.R1R2}"

cluster.json

@@ -14,7 +14,7 @@
 	"khmer" :
 	{
 		"queue" : "highmem.q,maiage.q",
-		"cluster" : "-l h_vmem={MEM}G"
+		"cluster" : "-l big_mem"
 	},
 	"megahit" :
 	{

global.smk

@@ -12,3 +12,4 @@ rule all:
 include: "../workflow_metagenomics/quality.smk"
 include: "../workflow_metagenomics/preprocess.smk"
 include: "../workflow_metagenomics/kaiju.smk"
+include: "../workflow_metagenomics/assembly.smk"