chip-seq read extension (36bp->200bp)

[vanbug]

Hola!

I was wandering if there's any command line tools (bedtools, samtools) other than R packages, to extend the reads from the base length b/w 36-50 bp to 200bp strand specific. I dont want to use R, as I have to load the file in the environment which is time consuming, I can make awk script easily, but there might be something present already. I want to do it because if I generate a coverage(bedGraph file) directly collapsing the bam file, I get small continuous mountains, which are more or less the referring to single gene target. (Look at the attached picture)

Also, another question is, does the chip-seq data should be normalized (same number of reads in control and sample) before calling peaks in Macs. For me, the mockIP control always less than 2/3rd the number of reads of control. I did a test, the number of positive peaks went down by 30-40% and -ve peaks came up by 30% after using Macs on normalized data.

Thanks
Sukhi

[ETHANol]

Most genome browsers don't like it when your reads extend beyond the end of chromosomes which will get a lot more of when you extend your reads. See:

http://ethanomics.wordpress.com/2011/11/06/bedendrepair-pl/

The second script does exactly what you are asking for.

[vanbug]

Thanks for the reply.
There is ucsc tool called bedclip which sorts the read off-end of chromosome problem.

I was tried using your script but I am getting "Unrecognized character \xE2" problems while executing it, I checked it and I think its because of utf encoding problems. Would providing a download link for the original source file, solve the problem!!.

Also, in this script do reconstruct the bed file after extension because the some reads in coverage will overlap. I am working with bedGraphs not bed. What do you think.

Eg:
Chr1 40 50 4
Chr2 110 150 3


After adding 150 to both co-ordinates, the reads will overlap and should be summed up (4th column-density), if I am not wrong.

Thanks

[ETHANol]

I was looking all over for a script that did what bedclip did. Thanks for the heads up. Anyway, it was a good motivation to learn some Perl.

It seems hosting code on wordpress.com is not trivial. I uploaded the file here:

https://sourceforge.net/projects/ethanomics/files/

[ETHANol]

It appears there is a UCSC utility that does what you are looking for as well called bedExtendRanges.

[vanbug]

Sorted

Hey,

Your perl script works fine for me , but the chromosome off-end clipping seems not to be the perfect one, but bedclip works fine as well.
I tried using bedExtendRanges but needs some database configuration, so dropped it.
So, I am currently using your script which extends the reads directionally, then bedClip to clip the off-end chromosome co-ordinates and mergeBed from bedTools to merge the overlapping reads.

Works good.

Cheers
Sukhdeep Singh