thanks for your answer and confirming what I suspected. Btw, in my case, at least in bowtie2, MAPQ contributes not much value (many alignments get the same value).
Hi,
SAM file specification has fields/tags such as alignment score (AS:i:<n> in bowtie2) and edit-distance (NM:i:<n>) for each end of the pair.
I am working with amplicons paired-end data (thus overlap is not uncommon) and would like to know if there is a paired-end "combined" score for the whole mate/pair (taking into account the overlap), e.g. for alignment score, edit distance, etc; this will be useful for later filtering etc.
I'm currently using bowtie2, and have just noticed "The alignment score for a paired-end alignment equals the sum of the alignment scores of the individual mates." written in the manual. Is that indeed the case especially when there is an overlap? How are the scores calculated for each end (0.5 for each ?)
I'm currently using my own implementation as a post-processing which is inefficient.
many thanks Shim
1 answer
Overlaps are not in any way taken into account when calculating the AS tag in bowtie2. There is no score reported that's a direct sum of the mate's AS score, though one can sort of use the MAPQ as an ersatz value for this, since MAPQ is dependent on the summed AS from the mates (as well as their summed XS values). But really if you want to compute a joint AS and such while accounting for overlaps then post-processing is your only option.
Log in to answer this question.