Add serovar detector

src/mmaseq/config/species_configs/A_pleuropneumoniae.yaml

@@ -0,0 +1,2 @@
+serovar_detector:
+    reads: True

src/mmaseq/config/species_configs/test.yaml

@@ -98,3 +98,6 @@ sistr:
 kmeraligner:
     database : [vancomycin, vancomycinOperon]
     reads: True
+
+serovar_detector:
+    reads: True

src/mmaseq/data/samplesheet.tsv

@@ -6,3 +6,4 @@ SRR26205262	reads/SRR26205262_1.fastq.gz	reads/SRR26205262_2.fastq.gz	assemblies
 ERR2929615	reads/ERR2929615_1.fastq.gz	reads/ERR2929615_2.fastq.gz	assemblies/ERR2929615.fasta	E_Faecalis.yaml
 ERR2929578	reads/ERR2929578_1.fastq.gz	reads/ERR2929578_2.fastq.gz	assemblies/ERR2929578.fasta	E_Faecium.yaml
 ERR142064	reads/ERR142064_1.fastq.gz	reads/ERR142064_2.fastq.gz	assemblies/ERR142064.fasta	C_difficile.yaml
+ERR14229029	reads/ERR14229029_1.fastq.gz	reads/ERR14229029_2.fastq.gz	assemblies/ERR14229029.fasta	A_pleuropneumoniae.yaml

src/mmaseq/workflow/envs/serovar_detector.yaml

@@ -0,0 +1,3 @@
+name: serovar_detector
+dependencies:
+  - bioconda::serovar_detector

src/mmaseq/workflow/rules/Analysis.smk

@@ -449,6 +449,33 @@ rule serotypefinder:
         mv $OUTDIR/{params.tmp_results} {output.results} >> {log.stdout} 2>&1
         """
 
+
+rule serovar_detector:
+    input:
+        assembly = rules.assembly.output.assembly
+    params:
+        tmp_results = "serovars.tsv"
+    output:
+        results = f"{outdir}/{{sample}}/raw/serovar_detector/serovar_detector.tsv"
+    conda:
+        ENVS_DIR / "serovar_detector.yaml"
+    log:
+        stdout = f"{logdir}/serovar_detector_{{sample}}.log"
+    shell:
+        """
+        INDIR = $(dirname {input.assembly})
+        OUTDIR = $(dirname {output.results})
+
+        cmd="serovar_detector -a $INDIR -o $OUTDIR -t 1"
+
+        echo "Executing command:\n$cmd\n" > {log.stdout} 2>&1
+        eval $cmd >> {log.stdout} 2>&1
+
+        echo "Renaming result files" >> {log.stdout} 2>&1
+        mv $OUTDIR/{params.tmp_results} {output.results} >> {log.stdout} 2>&1
+        """
+
+
 ### SNP Analysis ###
 
 # Samtools and bcftools

src/mmaseq/workflow/rules/Paired_Reads.smk

@@ -385,6 +385,34 @@ rule PR_serotypefinder:
         mv $OUTDIR/{params.tmp_results} {output.results} >> {log.stdout} 2>&1
         """
 
+
+rule PR_serovar_detector:
+    input:
+        R1 = lambda wc: samplesheet.loc[wc.sample, "read1"],
+        R2 = lambda wc: samplesheet.loc[wc.sample, "read2"]
+    params:
+        tmp_results = "serovars.tsv"
+    output:
+        results = f"{outdir}/{{sample}}/raw/PR/serovar_detector/serovar_detector.tsv"
+    conda:
+        ENVS_DIR / "serovar_detector.yaml"
+    log:
+        stdout = f"{logdir}/PR/serovar_detector_{{sample}}.log"
+    shell:
+        """
+        INDIR = $(dirname {input.R1})
+        OUTDIR = $(dirname {output.results})
+
+        cmd="serovar_detector -r $INDIR -o $OUTDIR -t 1"
+
+        echo "Executing command:\n$cmd\n" > {log.stdout} 2>&1
+        eval $cmd >> {log.stdout} 2>&1
+
+        echo "Renaming result files" >> {log.stdout} 2>&1
+        mv $OUTDIR/{params.tmp_results} {output.results} >> {log.stdout} 2>&1
+        """
+
+
 rule PR_lrefinder:
     input:
         res = rules.PR_kmeraligner.output.results,