Proper Bam / Tophat Output For Strand Specific Paired-End Data?

Question

[user]

If I have a strand-specific paired-end library and align with Tophat with an option like fr-firstrand, what is the proper to encode the strand of the pairs in the BAM, and how do mappers like Tophat handle it? With fr-firstrand we know that the first strand is the strand of the read, but by definition the second mate will be on opposite orientation. If the first read in the pair has strand + in its alignment in the BAM file, then will the second read in the pair also have strand + or can we count on it being -? Thanks

Answer 1

[Ido Tamir]

Tophat1 should give you:

  +/- or -/+ if you specify fr-X: --> <-- 
  +/+ or -/- if you specify ff-X: --> -->

Tophat1 also gives you singletons as output (i.e. only the first or second read map). Tophat2 can also give your discordant reads i.e. +/+ although fr (--> <--) was specified.

Important is the XS flag which tells you which strand the read comes from depending on X-firststrand or X-secondstrand.