Are velvet 'auto' setting parameters reliable?

[Dagga]

Hi,


I have just begun assembling genomes with velvet. I am new to de novo assembly and have been told the "auto" setting for -cov_cutoff and -exp_cov parameters are reliable for use. However, I have had some issues with auto setting for insert length.

Therefore, I was wondering if veteran velvet users have found that the auto setting for -cov_cutoff and -exp_cov are reliable or should I really spend the extra effort to work out the proper values for these parameters.

Cheers

[mastal]

I think the auto setting for -exp_cov is calculated using only the reads velvet uses in the assembly, so if you try to calculate a value for coverage based on the expected genome length and the number of bases in all your reads, you will end up with different values. Note that velvet uses kmer coverage.

As for -cov_cutoff, see the discussion in the velvet manual, you may want to try different cutoffs, and see what effect this has on your assembly. The -cov_cutoff calculated by velvet is half of the -exp_cov value.