Bwa Mapping Of Solid Paired End Reads

Question

[William]

Is it possible to use BWA to map Solid paired end reads , including the usage of the BWA Sampe pairing module? With BWA I mean the last version of BWA that supports colorspace, version 0.5.9.

What I understand now is that you can map the Solid PE reads and use bwa Sampe to pair the them, but the read pairs will be marked in the resulting BAM as not properly paired.

This is because BWA expects the reads to be in F3/R3 orientation, which they are not. Is the not properly paired bit the only side effect?

I read somewhere that the BWA Sampe mate rescue algorithm is hard coded with the expected orientation. Does it just not attempt to rescue missing mates when given pairs that are not in the expected orientation? Or does is introduce false positives (looking for the missing mate in the wrong direction)?

Is is true that you can just fix the not properly paired bit afterwards with Picard FixMateInformation ? Or are you left with an effect on downstream analysis tools that use the pairing information?

I looked at alternative mappers but I am happy with BWA for single fragment mapping and no other mappers suit al my requirements.

Answer 1

[Ian]

This is not quite the answer to your question, but I have found handling color-space paired-end read data from SOLiD very tricky. I had F3 (50bp) and F5-BC (35bp) data. In the end i could only get reasonable mapped reads using the BFAST-BWA program by Nils homer.

BFAST on its own would not map the F5 reads. BWA on its own did not make sense to me, however seemed to work quite well via BFAST. If oyu do use this approach inspect the expected parameters very carefully.

I hope this helps!