Finding Common Reads Across Multiple Fastq Files

Question

[Abhi]

Hi All

We have some metagenome samples(multiple illumina lanes). What I would like to do is find out % of reads that are common amongst these fastq's allowing upto #N mismatches.

I think I can take a subsample of the reads from each fastq/bin and compare them but just wondering if there is a slick approach to do the comparison.

Thanks! -Abhi

[Manu Prestat]

Do you allow difference of quality?

[Abhi]

@Manu : For now I dint think about it. I was just wondering if we can comapre the reads at base level and allowing 2-4 mismatches between the reads should cover for difference in quality scores.

Answer 1

[Mikael Huss]

I'd start by looking at the tools contained in vmatch. There are probably many ways to approach this problem but it seems sensible to use some kind of indexing on the fastq files prior to doing the comparisons.

[Istvan Albert]

neat software, did not about it before

Answer 2

[Manu Prestat]

cd-hit would do what you want. I just learnt by the way that you can directly use fastq file as input. You can also take a look at uclust (usearch).