How to Improve Newbler Assembly

[shuang]

I'm assembling 454 of a 160 KB fragment containing highly repeated sequences. This fragment is derived majorly from a centimere region.

I'm using Newbler to assemble this 454 sequencing data. I've played different combinations of 'min overlap length’ (40, 35) and ‘min overlap identity’ (90, 95, 98). By far, I only can get 22-25 contigs and 20kb-21kb N50.

I'm eager to know what other parameters and numbers could improve the assembly outcome.

[flxlex]

I don't think tweaking newbler parameters is going to help much. Perhaps other programs (e.g. celera, mira, clcbio) could do a better job. But you will gain the most from more data, in particular paired end (mater pair) reads, or the longer 454 reads that just became available or of course PacBio reads...).

[Kirstin]

How did you modify the minimum identity? i'm trying to do that in GS runMapping, but i'm finding that what I think should do that "-mi <percentage>" doesn't make any difference if I change the percentage? Even if I pick "-mi 99" it seems to include reads which are a much lower ID?