Backspaces in Chip-Seq data file

[mb492]

Hi,

I am new to NGS data analysis and I have got some ELAND output files (specificially the sorted.txt file) which I am planning to analyse using MACS. However, MACS keeps falling over due to a "Strand information can not be recognized in this line" error. I have deduced that this is due to a backspace characters which have appeared between some characters in my file and because MACS can't find a tab between the characters it complains that the line is not in the correct format.

Here is the offending line: (see the '^H' between the 0 and 1)

HWI-EAS486 23 1 97 11471 15019 0^H1 CAGGGTCACCCAGAGTGAGTGTGAAGCCAGCCTGAGATC hhYghhhhhhhggfhhghhhgghghghhhghghhhdfch chr10.fa 80424503 F 34G1C1A 6

Here is the same line as output by the MACS error: (backspace represented as x08 (HEX I think)
HWI-EAS486\t23\t1\t97\t11471\t15019\t0\x081\tCAGGGTCACCCAGAGTGAGTGTGAAGCCAGCCTGAGATC\thhYghhhhhhhggfhhghhhgghghghhhghghhhdfch\tchr10.fa\t\t80424503\tF\t34G1C1A\t6","34G1C1A


Does anyone have any idea how to replace these ^H (x08) backspace characters with tabs? the problem I have is that there are numerous occurances of ^H in the file which are legitimate.

Any help of advice would be very useful.

Thanks

[dpryan]

regular expressions

I would assume that you're running into a software bug in ELAND. I've only used data coming from it a couple times and don't recall ever running into backspaces.

Anyway, you could just remove the backspaces with regular expressions in your preferred programming language (or even perl from the command line, if you're into that sort of thing).

For example, in python, something like the following would replace all backspaces with tabs. A small amount of editing would restrict it to only the ones you want.

import re

f = open("your_file.txt","r")
of = open("a_new_file.txt","w")
for line in f:
    of.write(re.sub('\b','\t',line))
f.close()
of.close()

Assuming that the backspaces are only ever replacing a single character and you know (or can look up the ELAND file format), you could instead just use regex to parse the various fields:

re.search("(HWI\-EAS[\d]+)\.{1}([\d]+)\.{1}([\d]+)\.{1}([\d]+)\.{1}([\d]+)\.{1}([\d]+)\.{1}([\d]+)\.{1}([\d]+)\.{1}([ACGT]+)...",line)

Or something along those lines. There are a lot of ways one could do that. If you're uncomfortable doing that sort of thing yourself then you can probably find someone to write a short script for you in return for a pint or two of decent beer.