update HMM virulence prediction report

config.yaml

@@ -2,7 +2,7 @@ pathofact:
   sample: ["test_sample"] # requires user input
   project: PathoFact_update_trial # requires user input
   datadir: ../test_dataset # requires user input
-  workflow: "Tox" 
+  workflow: "Vir" 
   size_fasta: 1000
   scripts: "scripts"
   signalp: "/work/projects/ecosystem_biology/local_tools/SignalP/signalp-5.0b/bin" # requires user input

rules/Virulence/Combine_Virulence_SignalP.smk

@@ -10,7 +10,7 @@ import os
 rule merge_SignalPVir:
     input:
         Vir="{datadir}/{project}/VIRULENCE/HMM_classifier_virulence/{sample}_virulence_final_prediction.tsv",
-        SignalP="{datadir}/{project}/SignalP/aggregated/{sample}_SignalP_results.txt"
+        SignalP="{datadir}/{project}/SignalP/aggregated/{sample}_SignalP_results.tsv"
     output:
         "{datadir}/{project}/VIRULENCE/virulence_merged/{sample}_virulence_SignalP_prediction.tsv"
     params:

rules/Virulence/Virulence.smk

@@ -88,44 +88,23 @@ rule HMM_R_VIR:
         "../../scripts/hmm.R"
 
 # Give pathogenicity prediction
-rule HMM_VIR_nonpath:
+rule HMM_VIR_classification:
     input:
         "{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}.hmm_results.csv"
     output:
-        temp("{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}_hmm_non_path.txt")
+        non_path=temp("{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}_hmm_non_path.txt"),
+        pathogenic=temp("{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}_hmm_pathogenic.txt"),
+        unclassified=temp("{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}_hmm_unclassified.txt")
     params:
         runtime=config["pathofact"]["runtime"]["short"],
         mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"]
     shell:
         """
-        awk '$2 == "True" && $3 == "False"  && $4 == "False"' {input} | awk '$5 = "Non-Pathogenic"' | sed 's/ /\\t/g' > {output}
+        awk '$2 == "True" && $3 == "False"  && $4 == "False"' {input} | awk '$5 = "Non-Pathogenic"' | sed 's/ /\\t/g' > {output.non_path}
+        awk '$3 == "True"' {input} | awk '$5 = "Pathogenic"' | sed 's/ /\\t/g' > {output.pathogenic}
+        awk '$3 == "False"  && $4 == "True"' {input} | awk '$5 = "Unclassified"' | sed 's/ /\\t/g' > {output.unclassified}
         """
 
-rule HMM_VIR_pathogenic:
-    input:
-        "{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}.hmm_results.csv"
-    output:
-        temp("{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}_hmm_pathogenic.txt")
-    params:
-        runtime=config["pathofact"]["runtime"]["short"],
-        mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"]
-    shell:
-        """
-        awk '$3 == "True"' {input} | awk '$5 = "Pathogenic"' | sed 's/ /\\t/g' > {output}
-        """
-
-rule HMM_VIR_unclassified:
-    input:
-        "{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}.hmm_results.csv"
-    output:
-        temp("{datadir}/{project}/VIRULENCE/HMM_virulence/{sample}_hmm_unclassified.txt")
-    params:
-        runtime=config["pathofact"]["runtime"]["short"],
-        mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"]
-    shell:
-        """
-        awk '$3 == "False"  && $4 == "True"' {input} | awk '$5 = "Unclassified"' | sed 's/ /\\t/g' > {output}
-        """
 
 rule HMM_VIR_report:
     input: