search

vastgroup / vast-tools

A toolset for profiling alternative splicing events in RNA-Seq data.
MIT License
78 stars 29 forks source link

diff error caused by IR NA reporting format? #18

Closed timbitz closed 10 years ago

timbitz commented 10 years ago

'./vast-tools diff -a tg3_sequence -b ctrl_sequence -i /home/blencowe/blencowe1/rweather/AS_PIPE_v3/vast-tools/Hsa/cell_cycle/HeLa/drug/INCLUSION_LEVELS_FULL-Hsa2.tab -o /home/blencowe/blencowe1/rweather/AS_PIPE_v3/vast-tools/Hsa/cell_cycle/HeLa/drug/ -r 0.95 -m 0.1 -v FALSE > /home/blencowe/blencowe1/rweather/AS_PIPE_v3/vast-tools/Hsa/cell_cycle/HeLa/drug/diff_output_tg3.txt

Error in if (all(totalFirst < (opt$minReads + opt$alpha + opt$beta)) || : missing value where TRUE/FALSE needed Calls: mclapply -> lapply -> FUN '

timbitz commented 10 years ago

Culprit:

ARMCX4 HsaINT0012606 chrX:100779420-100786630 7211 chrX:100779191-100779419=100786631-100788446:+ IR-S NA N,N,N,ne,1@NaN,NaN NA N,N,N,ne,1@NaN,NaN

mirimia commented 10 years ago

The "ne" means there are "no effective" mappable positions in at least one of the junctions (EI, IE, or EE). I'm not sure how the scripts deal with these in the end (assigning automatically NA\tN,N,N,ne,1 to all?). In principle, they should never have PIR and/or OK quality, so they could be fully removed, but I'm not sure. The historical explanation for these is that Nuno didn't remove those with no effective, and then were carried around.

I would discard those for diff, in any case.

On Sun, Jul 27, 2014 at 6:08 PM, Tim Sterne-Weiler <notifications@github.com

wrote:

Culprit:

ARMCX4 HsaINT0012606 chrX:100779420-100786630 7211 chrX:100779191-100779419=100786631-100788446:+ IR-S NA N,N,N,ne,1@NaN,NaN NA N,N,N,ne,1@NaN,NaN

— Reply to this email directly or view it on GitHub https://github.com/vastgroup/vast-tools/issues/18#issuecomment-50277659.

kcha commented 10 years ago

I'm not sure, but I think they are just given NA.

For cRPKM, I replaced "ne" with NA as discussed previously with @mirimia.

UBrau commented 10 years ago

The "ne" is carried over from the "legacy" coverage score tables that is produced by RI_MakeCoverageKey.pl. Ulrich

On 27/07/14 12:08, Tim Sterne-Weiler wrote:

Culprit:

ARMCX4 HsaINT0012606 chrX:100779420-100786630 7211 chrX:100779191-100779419=100786631-100788446:+ IR-S NA N,N,N,ne,1@NaN,NaN NA N,N,N,ne,1@NaN,NaN

— Reply to this email directly or view it on GitHub https://github.com/vastgroup/vast-tools/issues/18#issuecomment-50277659.