What's the possible causes of strand bias in exome sequencing?

Question

[wangyi2412]

The strand bias here means the genotypes inferred from info presented by the forward strand and the reverse strand disagrees. For example,

read depths data at a position after mapping could be

         forward           reverse
ref      20                33
alt      19                0

which shows strong strand bias.

My question is that how come this happen? I think there might be 4 possible causes:

1. the sample itself does not strickly base paired
2. exome capture bias (how could this happen?)
3. sequencing error(how could this happen?)
4. mapping error

Am I right? Any comments is welcome!

[Devon Ryan]

Possibility 1 is pretty low probability. You forgot #5, PCR duplication that didn't get marked.

Answer 1

[mikhail.shugay]

Have you seen this paper: http://www.biomedcentral.com/1471-2164/13/666? They argue that this is likely to happen at option #4, mainly because of interference between local realignment and BAQ steps. The primary cause of this could be a sequencing error.

This question was also discussed in the community (http://seqanswers.com/forums/showthread.php?t=9354), which argues that a specific CCGG motif causes bases to be skipped on one strand, but not the other. As far as I remember, Illumina is likely to produce Indels in Poly-G regions, and Indels are common case for incorrect variant calling.

PS It would be also very useful if you'll specify what sequencing system, read alignment software and variant caller you're using

[wangyi2412]

Thank you! Sorry I reply late cause I did not have the Internet access.

Yes, the CCGG motif is a good hint! I will read it. Thank you very much!