Is it possible to do a variant call on a BLAT output?

Question

[BlatUser007]

Hey, i aligned reads (20 - 80 nt length) to the human genome. I won't get hits at the variable regions of the genome (eg. p-arm chr 13, 14, 15 ...), whereas the other parts are uniform covered. I am using a strict filter, to get exact matches. Is it possible to do a variant call or something like that to be more sensitive to this regions? Is it possible to map against the SNP-variants?

[cschu181]

First, I assume your reads are Illumina? Is there a reason why you're BLATing such small reads and not use a read mapper (bwa/bowtie(2))? By limiting to exact matches, you will not get any variant position (since you're filtering them out). I'd use one of the read mappers I mentioned. Since these are typically used to find variants in reads, you should be able get something out of it. You could also leave out the exact match filter, but obviously then you need to take measures against false positives (ensure only trust alignments backed up by sufficient read coverage.)

[BlatUser007]

Yes illumina reads. I can adjusted blat more sensitive than bwa (tile size, step size, ...). I also run a less strict filter and toke alignments containing 0 to 1 mismatch. Allowing one mismatch to a 20 nt read, means an identity of 95 % and i get more false positive results ... that's the reason why i am searching for a more sensitive way.

Answer 1

[JC]

No directly, you will need to create some scripts to do that, but it's better to use standard tools and pipelines for variant calling.