Merge branch 'addon_kaiju'

add-on.smk

+rule select_small:
+	input:
+		"work/metaSPADES/{reads}/contigs.fasta"
+	output:
+		contigs = "work/addon/{reads}.small_contigs.fasta"
+	params:
+		min_len = config["CONTIGS_LEN"]
+	shell:
+		"conda activate seqkit-0.10.1 "
+		" ; "
+		"seqkit "
+		"seq "
+		"--max-len {params.min_len} "
+		"--out-file {output.contigs} "
+		"{input} "
+		" ; "
+		"conda deactivate"
+
+rule kaiju_small:
+	input:
+		"work/addon/{sample}.small_contigs.fasta"
+	output:
+		"work/addon/{sample}.small_contigs.kaijuNR"
+	threads:
+		8
+	shell:
+		"kaiju "
+		"-t /db/outils/kaiju/nr/nodes.dmp "
+		"-f /db/outils/kaiju/nr/kaiju_db_nr_euk.fmi "
+		"-i {input} "
+		"-o {output} "
+		"-z {threads}"
+
+rule select_unmapped:
+	input:
+		"work/bowtie/align-{sample}-{reads}.bam"
+	output:
+		bam = temp("work/addon/unmapped-{sample}-{reads}.bam"),
+		fastq = "work/addon/unmapped-{sample}-{reads}.fastq"
+	shell:
+		"samtools "
+		"view "
+		"-f 4 "
+		"-o {output.bam} "
+		"{input} "
+		" ; "
+		"bedtools "
+		"bamtofastq "
+		"-i {output.bam} "
+		"-fq {output.fastq} "
+
+rule kaiju_unmapped:
+	input:
+		"work/addon/unmapped-{sample}-{reads}.fastq"
+	output:
+		"work/addon/{sample}-{reads}.unmapped.kaijuNR"
+	threads:
+		8
+	shell:
+		"kaiju "
+		"-t /db/outils/kaiju/nr/nodes.dmp "
+		"-f /db/outils/kaiju/nr/kaiju_db_nr_euk.fmi "
+		"-i {input} "
+		"-o {output} "
+		"-z {threads}"
+
+rule kronaNames_addon:
+	input:
+		"work/addon/{sample}.{addon}.kaijuNR"
+	output:
+		"work/addon/{sample}.{addon}.taxNames.tsv"
+	shell:
+		"kaiju-addTaxonNames "
+		"-t /db/outils/kaiju/nr/nodes.dmp "
+		"-n /db/outils/kaiju/nr/names.dmp "
+		"-i {input} "
+		"-o {output} "
+		"-r superkingdom,phylum,order,class,family,genus,species"

assembly.smk

@@ -76,12 +76,13 @@ rule metaspades:
 		unpack(assembly_input)
 	output:
 		contigs = "work/metaSPADES/{reads}/{reads}.contigs.fasta",
+		all_contigs = "work/metaSPADES/{reads}/contigs.fasta",
 		done = "work/metaSPADES/{reads}/done"
 	threads:
 		20
 	params:
 		min_len = config["CONTIGS_LEN"],
-		mem_tot = 250,
+		mem_tot = 500,
 		input = lambda wildcards, input: "-1 %s -2 %s"%(" -1 ".join(input.R1), " -2 ".join(input.R2)) if set(['R1', 'R2']).issubset(input.keys()) else "--12 %s"%" --12 ".join(input.R1R2),
 		output = lambda wildcards, output: os.path.dirname(str(output.contigs))
 	shell:
@@ -101,7 +102,7 @@ rule metaspades:
 		"seq "
 		"--min-len {params.min_len} "
 		"--out-file {output.contigs} "
-		"work/metaSPADES/{wildcards.reads}/contigs.fasta "
+		"{output.all_contigs} "
 		" ; "
 		"conda deactivate"
 

cluster.json

@@ -23,8 +23,8 @@
 	},
 	"metaspades" :
 	{
-		"queue" : "highmem.q,maiage.q",
-		"cluster" : "-l h_vmem=12G"
+		"queue" : "highmem.q",
+		"cluster" : "-l h_vmem=25G"
 	},
 	"diamond" :
 	{

global.smk

@@ -23,6 +23,9 @@ rule all:
 		expand("report/count-contigs-catalogue-{sample}.tsv", sample=config["SAMPLES"]),
 		"report/diamond_nr_catalogue.tsv",
 		"report/contigs_catalogue-taxNames.tsv",
+		#add-on
+		expand("work/addon/{sample}.small_contigs.taxNames.tsv", sample=config["SAMPLES"]),
+		#expand("work/addon/catalogue.{sample}.unmapped.taxNames.tsv", sample=config["SAMPLES"]),
 
 
 include: "../workflow_metagenomics/quality.smk"
@@ -32,3 +35,4 @@ include: "../workflow_metagenomics/assembly.smk"
 include: "../workflow_metagenomics/annotation.smk"
 include: "../workflow_metagenomics/count.smk"
 include: "../workflow_metagenomics/catalogue.smk"
+include: "../workflow_metagenomics/add-on.smk"