On Donnerstag 21 Januar 2010 Mihaela Angelova wrote: > Thanks a lot for answering me. > Mira is still not in any Pass now. The log assembly is stuck at this : > > Now running threaded and partitioned skimmer with 272 partitions in 2 > threads: [0%] ....|.... [10%] ....|.... [20%] ....|.... [30%] ....|.... > [40%] ....|.... [50%] ....|.... [60%] ....|.... [70%] ....|.... [80%] > ....|.... [90%] ....|.... [100%] > > I have ~668.000 reads (from 454) and ~75.000.000 paired-end reads in total. Hi Mihaela, uh, 75m reads? If it's a bacterium, that's far to many. If it's something larger ... hmm, I don't know why it should take an eternity right there. If you are interested I can give you a version with a lot of debugging. It won't help you immediately, but at least I'd have an idea where the problem is. > I have noticed that a huge number of paired > ends were discarded because they were too short : > > AAA/2 too small even with paired end > > Is this because their ends might have been clipped. Yes, exactly. > In the project_int_clippings.0.txt there are messages like : > > proposed cutback 1b: left AAA/2 9 -> 24 > > If I use the noclipping command, would it make the run more accurate? I very, very strongly suggest to use the proposed end clipping (PEC) for data sets with a coverage >15x (I suppose that to be the case for yours). PEC is inanely effective in doing in pre-screen and clips extremely effectively without clipping to much. If you don't use PEC, the assemblies are worse in 95% of the time, especially with Solexa and 454 reads. The output log gives you an idea how much was really clipped. It looks like this (example for a project with 800k 454 reads): Looking for proposed cutbacks ... done. Performed clips: Num reads cliped left: 31737 Num reads cliped right: 131198 Num reads completely killed: 7205 Total bases clipped : 2236490 Should these numbers be extremely high for you, then one should ask why the dataset is like that. 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