Align Paired End Reads Using Blast

Question

[shl198]

Hi all, Has anyone align illumina paired end reads using BLAST, I used gsnap to do the alignment first, then use BLAST to align the reads which were not mapped by gsnap. It seems that BLAST can only align single end reads. I aligned the two files separately and got results. But I don't know exactly how to deal with those results of BLAST. 1. Should I include the paired reads from two files or include all the reads as results? 2. How to merge the results with the sam file? Because I want to do assembly next step, I want to merge the blast results with gsnap results. Any comments would be appreciated. Thanks.

Answer 1

[Istvan Albert]

Some clarifications may be in order.

Alignments are performed on single sequence (single end). The pairing information is unrelated to performing the alignment. Its purpose is to help make decisions later about the alignments and their relative positions.

Traditional pair-end information processing assumes semi-global type alignment. Otherwise the concept of template (fragment) size is ill defined. The concept of fragment size assumes that both reads were found in the genome and reflects the distance between them. Blast will create local alignments.

Long story short:

1. turning BLAST into SAM file is probably not good idea
2. mixing outputs of a semiglobal and local aligners into the same SAM file is probably not a good idea

If all you need is to filter for the reads that do have blast hits then you can format the blast output into a tabular output that contains the sequence and then assemble those reads. There are also many other alternatives.

But I would suggest to keep the two methods separately and augment one with the other but don't treat them the same.