NewFile: get_bed_file_lengths.sh - downloads bedfile and counts lines…

…; NewFile: getCenterPointFromBed.py - returns chromosome, strandedness and midpoint of bed interval from input bed file;

getCenterPointFromBed.py

@@ -0,0 +1,25 @@
+import sys
+
+def parseBed(fname):
+  data = []
+  with open(fname) as f:
+    data = [l.strip().split() for l in f]
+  return data
+
+def extractRelevantFields(bed_list):
+  """Extract fields: e[0] (chromosome), centerpoint of bed interval 
+      e[5] (strandedness), and return as
+     list of tuples."""
+  return [tuple([e[0], (int(e[1]) + int((int(e[2]) - int(e[1])) / 2)), e[5]]) for e in bed_list]
+
+def main():
+  if len(sys.argv) != 2:
+    print("Usage: <exe> <bed_file.bed>")
+    sys.exit(1)
+
+  center_points = extractRelevantFields(parseBed(sys.argv[1]))
+  with open(sys.argv[1][:-4]+".bcp", 'w') as f:
+    f.write("\n".join([",".join([c, str(cp), s]) for c, cp, s in center_points]))
+
+if __name__ == '__main__':
+  main()

get_bed_file_lengths.sh

@@ -0,0 +1,7 @@
+for base in `cat $1`; do 
+  file=$base.bed.gz
+  curl -O -L https://www.encodeproject.org/files/$base/@@download/$file
+  echo $file >> all_bed_lengths.txt
+  gunzip -c $file | wc -l >> all_bed_lengths.txt
+  rm $file
+done