Add inference and apply new naming standards

.gitlab-ci.yml

@@ -97,7 +97,7 @@ inferMetadata:
   stage: unit
   script:
   - for i in {"chr8","chr4","chrY"}; do
-    echo " tin.py -i /project/BICF/BICF_Core/shared/gudmap/test_data/bam/small/Q-Y5JA_1M.se.sorted.deduped.${i}.bam -r /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/bed/genome.bed; cat Q-Y5JA_1M.se.sorted.deduped.${i}.tin.xls | tr -s \"\\w\" \"\\t\" | grep -P \"\\t${i}\\t\";"; done | singularity run 'docker://bicf/rseqc3.0:2.0.0' parallel -j 20 -k > Q-Y5JA_1M.se.sorted.deduped.tin.xls
+    echo "tin.py -i /project/BICF/BICF_Core/shared/gudmap/test_data/bam/small/Q-Y5JA_1M.se.sorted.deduped.${i}.bam -r /project/BICF/BICF_Core/shared/gudmap/references/GRCh38.p12.v31/bed/genome.bed; cat Q-Y5JA_1M.se.sorted.deduped.${i}.tin.xls | tr -s \"\\w\" \"\\t\" | grep -P \"\\t${i}\\t\";"; done | singularity run 'docker://bicf/rseqc3.0:2.0.0' parallel -j 20 -k > Q-Y5JA_1M.se.sorted.deduped.tin.xls
   - pytest -m inferMetadata
 
 downsampleData:

workflow/conf/biohpc.config

@@ -3,43 +3,49 @@ process {
   queue = 'super'
   clusterOptions = '--hold'
 
-  withName:getBag {
+  withName: getBag {
     executor = 'local'
   }
-  withName:getData {
+  withName: getData {
     executor = 'local'
   }
-  withName:parseMetadata {
+  withName: parseMetadata {
     executor = 'local'
   }
+  withName: trimData {
+    queue = 'super'
+  }
   withName: getRefInfer {
     executor = 'local'
   }
-  withName:getRef {
+  withName: downsampleData {
     executor = 'local'
   }
-  withName:trimData {
+  withName: alignSampleData {
     queue = 'super'
   }
-  withName:downsampleData {
-    executor = 'local'
-  }
-  withName:alignSampleData {
+  withName: inferMetadata {
     queue = 'super'
   }
-  withName:alignData {
+  withName: getRef {
+    executor = 'local'
+  }
+  withName: alignData {
     queue = '256GB,256GBv1'
   }
   withName: dedupData {
     queue = 'super'
   }
-  withName:fastqc {
+  withName: countData {
     queue = 'super'
   }
-  withName:inferMetadata {
+  withName: makeBigWig {
     queue = 'super'
   }
-  withName: makeBigWig {
+  withName: fastqc {
+    queue = 'super'
+  }
+  withName: dataQC {
     queue = 'super'
   }
 }

workflow/nextflow.config

@@ -20,39 +20,42 @@ process {
   withName: parseMetadata {
     container = 'bicf/python3:1.3'
   }
-  withName: getRefInfer {
-    container = 'bicf/awscli:1.1'
-  }
-  withName: getRef {
-    container = 'bicf/awscli:1.1'
-  }
   withName: trimData {
     container = 'bicf/trimgalore:1.1'
   }
+  withName: getRefInfer {
+    container = 'bicf/awscli:1.1'
+  }
   withName: downsampleData {
     container = 'bicf/seqtk:2.0.0'
   }
   withName: alignSampleData {
     container = 'bicf/gudmaprbkaligner:2.0.0'
   }
+  withName: inferMetadata {
+    container = 'bicf/rseqc3.0:2.0.0'
+  }
+  withName: getRef {
+    container = 'bicf/awscli:1.1'
+  }
   withName: alignData {
     container = 'bicf/gudmaprbkaligner:2.0.0'
   }
   withName: dedupData {
     container = 'bicf/gudmaprbkdedup:2.0.0'
   }
+  withName: countData {
+    container = 'bicf/subread2:2.0.0'
+  }
   withName: makeBigWig {
     container = 'bicf/deeptools3.3:2.0.0'
   }
   withName: fastqc {
     container = 'bicf/fastqc:2.0.0'
   }
-  withName:inferMetadata{
+  withName: dataQC {
     container = 'bicf/rseqc3.0:2.0.0'
   }
-  withName: makeFeatureCounts {
-    container = 'bicf/subread2:2.0.0'
-  }
 }
 
 trace {

workflow/scripts/parseMeta.py

@@ -10,7 +10,6 @@ def get_args():
     parser.add_argument('-r', '--repRID',help="The replicate RID.",required=True)
     parser.add_argument('-m', '--metaFile',help="The metadata file to extract.",required=True)
     parser.add_argument('-p', '--parameter',help="The parameter to extract.",required=True)
-    parser.add_argument('-e', '--endsManual',help="The endness.",required=False)
     args = parser.parse_args()
     return args
 
@@ -56,12 +55,9 @@ def main():
     # Get strandedness metadata from 'Experiment Settings.csv'
     if (args.parameter == "stranded"):
         if (metaFile.Has_Strand_Specific_Information.unique() == "yes"):
-            if (args.endsManual=="se"):
-                stranded = "--rna-strandness F"
-            elif (args.endsManual=="pe"):
-                stranded = "--rna-strandness FR"
+            stranded = "stranded"
         elif (metaFile.Has_Strand_Specific_Information.unique() == "no"):
-            stranded = ""
+            stranded = "unstranded"
         else:
             print("Stranded metadata not match expected options: " + metaFile.Has_Strand_Specific_Information.unique())
             exit(1)