denovo assembly and best coverage

[mmmm]

what is the best coverage that would be suitable to carryout denovo assembly (velvet) on illumina resequencing reads- would 20x be ok?

[Brian Bushnell]

what is the best coverage that would be suitable to carryout denovo assembly (velvet) on illumina resequencing reads- would 20x be ok?

I've found Velvet to work best at around 40x, using 150bp reads. But it depends somewhat on the read length; with longer reads you may be able to get away with less coverage.

[mmmm]

thank you- I have 250PE and 300PE reads (but of lower coverage 20x) so will try velvet and see

I was wondering is it ok to combine fastq files of 250PE and 300PE redas (of the same sample) using cat command, would this combination affect the analysis (either velvet or bwa??

[mastal]

for velvet it might be better to leave the 2 sets of reads separate, in case the library prep gave a different insert length for the different runs. You can still use both sets of reads in the same run of velvet though, just flag one set as
'-shortPaired', and the second set as '-shortPaired2' when running velveth, but make sure you have velvet compiled with at least 'CATEGORIES=2'.

[mmmm]

thank you- will give it ago
so if I have 3 diff PE reads (for the same samples but from diff runs)- I will make CATEGORIES = 3

velveth output/ kmer -fastq -shortPaired -separate run1.read1.fastq run1.read2.fastq -shortPaired2 -sparate run2.read1.fastq run2.read2.fastq and so on?