update signalp rules

config.yaml

 pathofact:
   sample: ["test_sample"] # requires user input
-  project: PathoFact_update_trial # requires user input
+  project: PathoFact_update_trial_new # requires user input
   datadir: ../test_dataset # requires user input
   workflow: "complete" 
   size_fasta: 1000

rules/Universal/SignalP.smk

@@ -3,15 +3,31 @@
 import glob
 import os
 
+# Split sequences for signalp v5.0 (max. 5000 seq)
+checkpoint splittingsignalP:
+    input:
+        "{datadir}/{project}/renamed/{sample}_ID.faa"
+    output:
+        splits=directory("{datadir}/{project}/SignalP/splitted/{sample}_dir/")
+    params:
+        outdir="{datadir}",
+        runtime=config["pathofact"]["runtime"]["short"],
+        mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"],
+        split=config["pathofact"]["size_fasta"]
+    conda:                                          
+        "../../envs/Biopython.yaml"                                                
+    shell:                                                      
+        """                                                                 
+        python {config[pathofact][scripts]}/split.py {input} 2000 {wildcards.datadir}/{wildcards.project}/SignalP/splitted/{wildcards.sample}_dir
+        """
+
+
 #Run SignalP on split sequence files
-rule signalp:
+rule signalp_gramp:
     input:
-        "{datadir}/{project}/renamed/{sample}_ID.faa"        
+        "{datadir}/{project}/SignalP/splitted/{sample}_dir/{file_i}.fasta"        
     output:
-        SignalP_gramP="{datadir}/{project}/SignalP/{sample}/gramp_summary.signalp5",
-        SignalP_gramN="{datadir}/{project}/SignalP/{sample}/gramn_summary.signalp5"
-    log:
-        "{datadir}/{project}/SignalP/{sample}.log"
+        SignalP_gramP="{datadir}/{project}/SignalP/Gram+/{sample}/{file_i}_summary.signalp5"
     message:
         "Running SignalP analysis on {input} resulting in {output}."
     params:
@@ -21,10 +37,70 @@ rule signalp:
     shell:
         """
 	export PATH={config[pathofact][signalp]}:$PATH
-        signalp -fasta {input} -org gram+ -prefix {wildcards.datadir}/{wildcards.project}/SignalP/{wildcards.sample}/gramp -batch {config[pathofact][size_fasta]}
-        signalp -fasta {input} -org gram- -prefix {wildcards.datadir}/{wildcards.project}/SignalP/{wildcards.sample}/gramn -batch {config[pathofact][size_fasta]}
+        signalp -fasta {input} -org gram+ -prefix {wildcards.datadir}/{wildcards.project}/SignalP/Gram+/{wildcards.sample}/{wildcards.file_i}
         """
 
+rule signalp_gramn:
+    input:
+        "{datadir}/{project}/SignalP/splitted/{sample}_dir/{file_i}.fasta"
+    output:
+        SignalP_gramN="{datadir}/{project}/SignalP/Gram-/{sample}/{file_i}_summary.signalp5"
+    message:
+        "Running SignalP analysis on {input} resulting in {output}."
+    params:
+        outdir="{datadir}",
+        runtime=config["pathofact"]["runtime"]["long"],
+        mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"]
+    shell:
+        """
+        export PATH={config[pathofact][signalp]}:$PATH
+        signalp -fasta {input} -org gram- -prefix {wildcards.datadir}/{wildcards.project}/SignalP/Gram-/{wildcards.sample}/{wildcards.file_i}
+        """
+
+def aggregate_signalpP_input(wildcards):
+    checkpoint_output= checkpoints.splittingsignalP.get(**wildcards).output.splits
+    return expand(
+        "{datadir}/{project}/SignalP/Gram+/{sample}/{file_i}_summary.signalp5",
+        datadir=wildcards.datadir,
+        project=wildcards.project,
+        sample=wildcards.sample,
+        file_i=glob_wildcards(os.path.join(checkpoint_output, "{i}.fasta")).i
+    )
+
+def aggregate_signalpN_input(wildcards):
+    checkpoint_output= checkpoints.splittingsignalP.get(**wildcards).output.splits
+    return expand(
+        "{datadir}/{project}/SignalP/Gram-/{sample}/{file_i}_summary.signalp5",
+        datadir=wildcards.datadir,
+        project=wildcards.project,
+        sample=wildcards.sample,
+        file_i=glob_wildcards(os.path.join(checkpoint_output, "{i}.fasta")).i
+    )
+
+rule SignalPP_aggregate:
+    input:
+        aggregate_signalpP_input
+    output:
+        "{datadir}/{project}/SignalP/{sample}/gramp_summary.signalp5"
+    params:
+        outdir="{datadir}",
+        runtime=config["pathofact"]["runtime"]["short"],
+        mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"]
+    shell:
+        "cat {input} > {output}"
+
+rule SignalPN_aggregate:
+    input:
+        aggregate_signalpN_input
+    output:
+        "{datadir}/{project}/SignalP/{sample}/gramn_summary.signalp5"
+    params:
+        outdir="{datadir}",
+        runtime=config["pathofact"]["runtime"]["short"],
+        mem=config["pathofact"]["mem"]["normal_mem_per_core_gb"]
+    shell:
+        "cat {input} > {output}"
+
 rule aggregate_signalP:
     input:
         SignalP_gramP="{datadir}/{project}/SignalP/{sample}/gramp_summary.signalp5",