BWA mem and BWA aln lead to different results

Question

[zhangdezhi008]

Hi, all,

I tried two uncover the genetic structure for a bird population using resequencing data, with PAIRED-END 100 bp reads. I used two different mapping methods BWA mem and BWA aln, and then used SAMtools to get vcf file. After filtering, we get two different SNP datasets when using BWA mem and BWA aln, and of course, this leads to two diffrent genetic structures for the same population! The problem is the mapping methods BWA mem and BWA aln, which one should I trust?

Any suggestions would help a lot! My best wishes, Dezhi

[GenoMax]

Take a look at associated links: C: BWA aln vs. mem for 100bp reads

Answer 1

[dariober]

this leads to two diffrent genetic structures for the same population!

Provided that there are no bugs in your analysis pipeline, I think the question is not so much which aligner is better but rather on whether there is an actual population structure or whether the two different structures are actually the same.

I mean, bwa mem seems to be the standard now and it has superseded bwa aln (so use mem I would suggest). Still, bwa aln should give good alignments and if your final results depend on the choice between two good aligners, then these results are not very robust and should be taken with care,

[zhangdezhi008]

Thanks for your reply very much! According to the previous study, this species should have population structure with two lineages based on sanger sequences, which is consistent with the result generated from BWA-aln. But as you said, BWA-mem should be the better solution for mapping. Confused still...