How to selectively filter BLAST results for an endogenous retroviral LTR to retrieve members of the same ERV family?

Question

I'm running a local BLAST search on a HERV-K(HML2) LTR sequence in the human genome. I get thousands of hits.

I want to retrieve the hits that correspond to other HERV-K(HML2) LTR sequences only.

What alignment length/identity filters should I use on the hits to select LTRs of the same family? Insertions and deletions could easily alter the alignment lengths meaning the cutoff length for the filter I want isn't clear. I can work out what alignment length cutoffs remove what percentage of the results to identify big drops at specific cutoffs, but it hasn't been very helpful.

Answer 1

I don't think there is a definite answer to this problem, but here are a few things that might help you, apart from a simple length filter for the hits:

• Filter for regions: since LTRs are repeats you might get multiple BLAST results from the same repeat region/locus (especially if your input sequence is smaller than said region). You probably want to keep only one hit per region

• Check for a conserved (sub)sequence: hopefully your specific LTR sequence has a 'core' that needs to be present, so that it can be considered part of your familiy. You only want to keep BLAST hits that have an absolute match for this 'core'.

Answer 2

I don't know what your application is, but RepBase has a curated set of ERV loci in a BED file that you can use to retrieve sequence/define regions of interest et cetera.

Of course, generally speaking repetitive regions may be harder to map in terms of genomic location, and in my lab, where we regularly study ERVs in cancer, we usually map RNAseq data to Repbase sequences in a fasta file to derive total expression estimates on a per-family basis.