On Dienstag 04 August 2009 Joseph Fass wrote: > I'll chime in here, hoping this is relevant as well: Indeed it is. > Now, that's velvet, not Mira ... but, I'm wondering, is there a reason for > this from the field of information theory? Does using ultra-high coverage > ensure that there's an error of every kind at every positon, this confusing > even a "perfect" assembler? Can anyone comment on this? I can. After having seen the 100x coverage from Johann, I can say you've hit the nail on the head, at least in the case of his simulated data. More on that in a short while in a mail responding to Johann. For Solexa, things get more complicated: with the healthy dose of trimming MIRA applies to reads, you can go without problems to coverages in the 60s or more ... but then other aspects come into the game. Solexa reads have a weakness with the sequence GGC.G (5' -> 3'): bases called after this motif are, in some cases (probably due to other secondary structures forming) *very* error prone. Up to a point where enough errors accumulate at a single place which fool the trimming routines and subsequently trick MIRA to believe it's a repeat with slightly different bases ... and disassemble these things. Quite annoying. Regards, Bastien -- You have received this mail because you are subscribed to the mira_talk mailing list. For information on how to subscribe or unsubscribe, please visit http://www.chevreux.org/mira_mailinglists.html