in silico enrichment

Question

[kamhamea]

I'm looking for a fast tool to scan hight throughput sequencing data reads for patterns, so that the final thorough analysis can be performed with only a few reads that belong to the cluster.

Practicaly, I manually generated a set of oligos (~20mer) that uniquely belong to a cluster of genes, and now I'm going to find all the reads that are matching. Next step is finding neighboring reads, but already the fist step programmed based on a python find regex string routine takes weeks on whole genome seq.

Answer 1

[Brian Bushnell]

This sounds like a job for BBDuk, which can filter reads by matching kmers. It's extremely fast. For example, using 20-mers:

bbduk.sh in=reads.fq outm=matching.fq ref=oligos.fa k=20 mm=f

Note that if your oligos contain degenerate IUPAC symbols like "N" you should add the flag "copyundefined".