[wgs-assembler-users] upper limit on coverage in kmers

[Walenz, B. <bw...@jc...>]

Hi, Geoff-

In the context of merTrim, no.  It uses the mer database to find trusted
kmers.  It guesses the coverage contained in the input reads, then picks a
lower limit on what 'trusted' should be.  IIRC, trusted is 1/4 the input
coverage -- 15x of input reads will treat any kmer with less than 4
occurrences as erroneous.

In the context of overlaps, no.  It uses the mer database to avoid seeding
an overlap alignment on high-frequency seeds.

b


On 6/13/12 4:45 PM, "Waldbieser, Geoff" <Geo...@AR...>
wrote:

> When producing a mer database from a set of paired end Illumina reads, you
> recommend a lower limit of 15X genome coverage. Is there an upper limit to the
> amount of genome coverage so that one does not oversample sequencing error?
> 
> Geoff Waldbieser