Option to count partially mapped reads

In case more flexibility is required for counting.
Keep partial_map row for compatibility when comparing
with strictly mapped reads.

.test/test.config

@@ -41,6 +41,9 @@ params {
 
  // fasta file of guide sequences, leave blank to skip guide counting
  guides_fasta = "$projectDir/.test/data/guides_simulated.fasta"
+
+ // count partially mapped reads
+ lenient_counts = false
 }
 
 //SQL DB Plugin

bin/count_guides.py

@@ -32,6 +32,9 @@ def parse_args():
  parser.add_argument('guide_reference',
  type=str,
  help='Guide reference fasta file.')
+ parser.add_argument('--lenient',
+ action='store_true',
+ help='Count partial mappings.')
 
  return parser.parse_args()
 
@@ -69,13 +72,14 @@ def main():
  continue
 
  guide_len = guide_lens[read.reference_name]
- if read.reference_end - read.reference_start != guide_len:
+ read_spans_ref = read.reference_end - read.reference_start == guide_len
+ if args.lenient or read_spans_ref:
+ counts[read.reference_name] += 1
+ else:
  counts['partial_map'] += 1
  print(f'Guide {read.reference_name} does not span the whole guide',
  file=sys.stderr)
 
- counts[read.reference_name] += 1
-
  sample_name = os.path.basename(args.bam).split(".")[0]
  print(f'guide\t{sample_name}')
  for guide, count in counts.items():

modules/count.nf

@@ -42,6 +42,7 @@ process CountGuides {
 
  script:
  def outCounts = "${sampleName}_guide_counts.txt"
+ def lenientFlag = params.lenient_counts ? "--lenient" : ""
  /*
  count collation is a hacky bash script to get collated output,
  the script pastes the count files together, and then cuts out
@@ -58,7 +59,7 @@ process CountGuides {
  samtools view -S -b | \
  samtools sort -o \${sample}.bam
 
- count_guides.py \${sample}.bam ${params.guides_fasta} > \${sample}_counts.txt
+ count_guides.py \${sample}.bam ${params.guides_fasta} ${lenientFlag} > \${sample}_counts.txt
  done
 
  paste *_counts.txt > tmpfile

nextflow.config

@@ -41,6 +41,9 @@ params {
 
  // fasta file of guide sequences, leave blank to skip guide counting
  guides_fasta = ""
+
+ // count partially mapped reads
+ lenient_counts = false
 }
 
 //SQL DB Plugin

nextflow_schema.json

@@ -132,6 +132,11 @@
  "default": "",
  "fa_icon": "fas fa-file",
  "description": "Reference fasta file containing guide sequences (leave blank to skip guide counting)."
+ },
+ "lenient_counts" : {
+ "type": "boolean",
+ "default": false,
+ "description": "If true, count partially mapped reads."
  }
  }
  },