Gene mutations discovery

Question

[francesca.defilippis]

Hello!

I'm approaching a dna mutation analysis for the first time. I have DNA sequences of a structural gene (454 reads, about 450 bp, single end) sequenced in 10 samples and I need to find mutations and quantify the abundance of each mutant in each sample.

This is how I'm thinking to proceed:

• demultiplexing sequences using qiime
• trimming low quality bases and filtering reads too short
• clustering them at 100% similarity using usearch
• aligning centroids to some sequences found at NCBI using mummer

My questions are:

First of all, is my approach correct? Does anybody have experience in this kind of analysis and which approach is used?

Then, I don't know if mummer is the good choice... It is designed for snp discover in genomes, so I don't know if it could be a good idea using it for sequences that I expect to be very similar among them. If not, any other suggestion about software to use?

Thanks

Francesca

Answer 1

[Devon Ryan]

I'd skip clustering and just align everything. Just use a few more cores. I'd give BWA a try rather than mummer, mostly because BWA typically produces good quality results in downstream variant calling and also because it's reasonably fast.

[francesca.defilippis]

Hi! Thanks for your prompt reply! I wondered about using bwa or bowtie, but using them wouldn' t imply losing new variants? I could just quantify the number of sequences matching those I have as references. Is it right?

[Devon Ryan]

BWA is probably the most popular aligner used to find new variants, so no it won't lose them (unless the sequence is very highly divergent).