Query on K-mer using in velvet

[ramadatta.88]

Hi, I am newbie to NGS. TO assemble a genome i started using velvet. During which I came across a term called 'K-mer' which means the overlapping segment in the reads. I would like to know that if the K-mer length assigned is suppose 21, is it that the overlapping is done exactly for 21 nucleotides at the ends or is it a minimum requirement for the reads to match? Could i kindly know the difference..Thanks in advance..

[boetsie]

Hi!

A read is hashed into your defined k-mers. And only ends are considered of the k-mers, simply because you are stripping the read.

Say you have a 10bp read, and a k-mer of 5bp, the hashing of the read into kmers will be;

read: ACGATGATAGTA
kmer: ACGAT
kmer: .CGATG
kmer: ..GATGA
kmer: ...ATGAT
kmer: ....GATAG
kmer: .....ATAGT
kmer: ......TAGTA

The program than finds k-1 overlap (thus here: 5-1=4 bp overlap) at the ends of the k-mers.

E.g:
ACGAT
.CGATG


So you find overlap at the ends of the k-mers, but within the reads.

Have a look at this poster from Velvet for maybe detailed information;

http://www.ebi.ac.uk/~zerbino/velvet/velvet_poster.pdf

Regards,
Boetsie

[ramadatta.88]

thank you..My doubt is cleared..