[Denovoassembler-devel] RE : [PATCH] clean up / simplify CoverageDistribution function

[Sébastien B. <seb...@ul...>]

Send me your problematic phix CoverageDistribution.txt file and I will add it to my unit tests.

If coverage decreases sharply from the starts, chances are that it won't work with Ray.



> ________________________________________
> De : David Eccles (gringer) [dav...@mp...]
> Date d'envoi : 22 juillet 2011 07:13
> À : den...@li...
> Objet : [Denovoassembler-devel] [PATCH] clean up / simplify     CoverageDistribution function
> 
> I've been trying to clean up the coverage distribution function to make
> it easier for me to understand.
> 
> Using this patch, it works with my simulated phiX data (because of the
> fallback to max coverage with no votes -- it looked like this was in the
> code before?).
> 
> However, it fails for the S.med 454 data (transcriptome data, in which
> coverage values decrease sharply from the start). I noticed that I can
> 'fix' the peak finding code by making this change (about line 83):
> 
> if((votes[i] > votes[largestPosition])
> ->
> if((votes[i] >= votes[largestPosition])
> 
> But then the minimum position is set to the largest position, and the
> assembler panics. I need to set the minimum coverage to the coverage at
> the first position for things to work with that data:
> 
> m_minimumCoverage=x[0]; // about line 98
> 
> I haven't put these additional changes in the patch, because I'm not
> sure how they will affect other things... and [as I've just noticed] I
> get a segfault on the 454 data with my code when doing this.... I'll try
> to nail that down in the next day or so.
> 
> -- David
> 
                  Sébastien Boisvert
                  http://github.com/sebhtml/ray