sam/bam to wig for rna-seq

[biocomputer]

What is the best way to convert rna-seq sam/bam files to .wig files? I've come across a few possibilities: deeptools bamcoverage, samtools pileup, samtools depth, bedtools genomecov, igvtools count. Can any of these properly deal with the stranded, spliced, paired-end reads usually generated by rna-seq?

[dpryan]

Do you want separate files for each strand or are you ok with both strands being merged? If you're ok with merged values, then samtools pileup (or even just depth) would suffice.

[biocomputer]

I would like them separate so I can differentiate transcription between the plus and minus strands.

[dpryan]

If you can't find a prewritten tool then you can do this with either pysam using a pileup or the htslib C API. If you're familiar with either of those then this won't be too bade to write, but if not it'll be a bit daunting. If I have a chance this weekend then I'll see if I can just quickly code something, though the odds of me having a half hour to write and test something before Monday is relatively low.

[blancha]

1. Split the BAM files by strand with samtools view (& samtools merge).
2. Generate bedgraph files for the positive strand and negative strand BAM files with bedtools genome cov.
3. Generate bigWig files with bedGraphToBigWig

[dpryan]

Since I can foresee needing something like this at some point anyway, here's a quick C implementation using htslib (the basis for samtools): https://github.com/dpryan79/Answers/...Qanswers_48599

If you clone that repo, switch to the SEQanswers_48599 directory and type "make", you'll get a bam2wig program that should do what you want. You can filter by MAPQ and phred scores if you want and you should be able to set how the strand is determined (the default is for dUTP-based libraries, which are the most common). This has not been tested other to ensure that it actually runs, so have a look at the output in IGV and ensure it matches the raw data.

[biocomputer]

Thank you for the replies, I'm going to try both of the options given here and a few others and see what's the best.

[ptr]

Originally Posted by dpryan

Since I can foresee needing something like this at some point anyway, here's a quick C implementation using htslib (the basis for samtools): https://github.com/dpryan79/Answers/...Qanswers_48599

Hi Devon,

thanks for the code, I used it as a template for my own implementation.
I have a question regarding your -p option for the min base quality of a base being counted: You write that 0 would indicate overlapping bases from paired-end reads, which should be not counted twice. I am wondering if there is a mapper that sets those bases to 0 upon encountering overlapping mates while mapping? Or is this done by on a read merging program prior to mapping?

thanks,
Peter

[dpryan]

Hi Peter,

I'm not aware of any aligners that do that for you, this is usually done post-hoc...though honestly it'd be an easy enough thing to do. This is only really needed when people are making pileups, at which point it's easy enough to just do it then.

[Richard Finney]

I like this bam2wig script ... just need samtools and awk ...

samtools depth -aa your.bam | awk '{if (prev!=$1) print "fixedStep chrom="$1" start="$2" step=1 span=1\n"$3; else print $3; prev=$1}' > your.wig