disulfidebond/parse_bed_file.md

## parse_bed_file.md

      
    Raw
  

              parse_bed_file.md
            
          
    Overview

A BED file needs to be parsed and reformatted into a CSV file. Broadly speaking, there are two options: use a GUI, or use scripting.
GUI

Use BBEdit, Atom, or Apple's TextEdit (see caution) to search and replace. An example is shown here:


Caution: Apple's TextEdit has an extremely useful GUI for search and replace, that even simplifies Regex. However, it may replace some characters with one that is not recognized by all text editors, such as the double-quote character. You've been warned.

Scripting

I haven't been able to find a one-step process, but the following scripts accomplish the task, and also can serve as a cookbook for similar tasks.


Parse out only the track line and the following line:
  grep -A1 'track' radCohort_geneList.bed.txt > radCohort_geneList.unparsed.txt 
  # truncated output:
  # track name="SOMETHING" description=""
  # chr1  87863625	87864548 
  # --


remove grep characters
  perl -pe 's/--\n//' radCohort_geneList.unparsed.txt > tmp


append bed line to track line. Use a Regex negative lookahead
  perl -0pe 's/\n(?!([a-z]{5}|$))//g' tmp > tmp2
  # truncated output:
  # track name="ATM" description="ATM serine/threonine kinase [Source:HGNC Symbol;Acc:HGNC:795]" chr11      108222484 


reformat line into comma-separated and parse out bed entries
  perl -pe 's/\t/,/g' tmp2 | cut -d, -f1 > tmp3
  # commands can be piped into each other, example:
  perl -pe 's/\t/,/g' tmp2 | sed 's/itemRgb=\"On\"/chromosome=\"/g' | cut -d, -f1 | sed 's/$/\"/' > tmp4
  # truncated output from the above piped command is:
  # track name="ATM" description="ATM serine/threonine kinase [Source:HGNC Symbol;Acc:HGNC:795]" chromosome="chr11"


reformat space-delimited into a comma-separated file, in this case the pattern '" ' is used, but this can be modified as necessary
  sed 's/\" /,/g' tmp4 > radCohort_geneList.txt


finally, remove the 'track ' string from the beginning of the line:
  sed 's/^track //g' radCohort_geneList.txt > radCohort_geneList.csv