single gene denovo assembly?

[asunen]

Hello,

I'm looking at population level variation in a 100kb region in Drosophila. When I use standard alignment programs (bwa and bowtie) on reads from some sequenced populations, I keep losing insertions that I know are present (and appear in the raw reads), but are missing in the reference genome.

Someone suggested I do a denovo assembly on my 100kb region (since I'm not really interested in the rest of the genome), to get around this problem.

Is there a way to filter reads for a specific region using denovo assembly? Is this a good approach to the problem, or am I better off tweaking my methods with alignment to the reference genome?

Thanks!

[jgibbons1]

I've actually done something similar. I had whole genome data but for a particular piece of analysis I was only interested in a highly variable 35kb region. I also knew that I was missing indels in mapping results.

I ended up compiling a database with this 35kb region (plus 5kb flanking on each side) from all of the sequenced genomes/genes in my region I could find. I then blasted my reads against this database. Reads which significantly hit my region were extracted and then independently de novo assembled. It worked very well.

Hope this helps.

John

[asunen]

Thank you, I'll try that

[swbarnes2]

The only issue is, how do you know which reads are in your region? If your sample has an indel as compared to the reference, the reads might not map there, so you won't know to include in the de novo. You'd have to denovo everything, or do another experiment where you long PCR around your region of interest.