View Single Post
Old 03-15-2011, 04:59 PM   #3
swbarnes2
Senior Member
 
Location: San Diego

Join Date: May 2008
Posts: 912
Default

Building your sequence de novo generally requires higher coverage, and you'll end up with lots of contigs that can't be bridged due to repetitive or non-complex sequence. Aligning to a very closely related genome will allow you to easily find discrepancies, even if coverage is not so great. If your coverage is, say, an even 5x, you could call a lot of SNPs. I've found plenty of real, sanger-verified SNPs with 3x coverage. But no de novo assembler will do anything with that.

Also, for a lot of applications, what you really want are the discrepancies between your sample and a reference, so aligning to that reference is a much easier way to find them, rather than building a de novo sample reference, and comparing that to something. Also, de novo won't be totally accurate if your sample is heterozygous, or a mix of samples; at least not Velvet, which is a popular one, and the one I'm most familiar with. Whereas with an aligner, it isn't hard to see that so many reads show one allele at a locus, and so many reads show another.
swbarnes2 is offline   Reply With Quote