kaiju : small contigs + unmapped

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"

global.smk

@@ -22,6 +22,9 @@ 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"
@@ -31,3 +34,4 @@ 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"