Velvet - stats output - min_contig_length

[ytmnd85]

I am attempting to use Velvet to analyze some data from our lab. During a practice run, I noticed that the output from velvet had some odd contigs (length 1, coverage 300,000), so I decided to place some parameters on it. I set the minimum coverage to 10, and the minimum contig length to 10. When I checked the new output, the minimum coverage had been implemented, but it didn't appear that the minimum contig length had. Looking at the stats file, I still had contigs of length 1. However, upon looking at the contigs.fa file, I found that my contig length had been implemented.

Does anyone know why my contig file is correct, but my stats file does not reflect my contig-length restriction?

[Zigster]

Does anyone know why my contig file is correct, but my stats file does not reflect my contig-length restriction?

yeah that parameter only affects the contigs.fa file (as you have already discovered)

min_contig_length is just a dumb cutoff included for convenience - a two-line perl script could accomplish the same thing

cov_cutoff is a real parameter that can actually change the assembly in unexpected ways

and also note that length in the stats.txt file is in kmers, not bp

[ytmnd85]

Thank you,
That makes sense now. It just had me perplexed when I was trying to interpret my results and I was seeing contradicting files.

[mmmm]

kmer to nucleotide

yeah that parameter only affects the contigs.fa file (as you have already discovered)

min_contig_length is just a dumb cutoff included for convenience - a two-line perl script could accomplish the same thing

cov_cutoff is a real parameter that can actually change the assembly in unexpected ways

and also note that length in the stats.txt file is in kmers, not bp

in stats.txt (velvet output) how to convert node length to the actual nucleotide length, for example, if the node length is 1what would be the nucleotide length?

[mastal]

see the equation for calculating kmer coverage from nucleotide coverage in the velvet manual.

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

"The relation between k-mer coverage Ck and standard (nucleotide-wise) coverage C is Ck = C (L−k +1)/L where
k is your hash length, and L you read length."


Node length = Contig length in nucleotides minus k-mer length + 1.