Could direction of reads affect the mapping of medium-size insertions

Question

[s.maxouri]

Dear all,

I'm sequencing samples with S5 Ion torrent and I'm experiencing some trouble with the detection of Insertions (~20bp). For the mapping I'm using tmap. I'm detecting an insertion of 19bp (to be more precise, a region of 19bp is duplicated) and this INS is evident only in the forward strand. Due to the weakness of tmap to detect efficiently INDELS, I also used bbmap and in the resulting BAM the insertion is detected in both reverse and forward strand. I have come across this issue in several different cases (different samples and genes) and in all of them, when using the tmap, insertion is detected only in one direction.

My question is could the direction of reads (forward or reverse) and the flanking regions (upstream and downstream the insertion) affect the mapping process and result in such a strand bias?

Thank you in advance, Stella

Answer 1

[Istvan Albert]

TMAP is really old software, the last commit was almost 10 years ago

https://github.com/iontorrent/TMAP

don't use it. Use bbmap, bwa or something that actually works.

[s.maxouri]

I know, but unfortunately I'm forced to use this software for a series of reasons. This is the reason I want to verify if these insertions (with high strand bias) are real or artifact, also to understand the source of the problematic detection in the case of tmap.

[Istvan Albert]

it seems that the tool does not produce the proper answer, it is both obsolete and likely incorrect -

it would be interesting to know - if for the selfish reasons of satisfying my morbid curiosity: How come you are forced to use it? Who or what circumstance demands using an obsolete tool that produces incorrect results? Why would anyone want to do that?