Thread: subread
View Single Post
Old 06-03-2013, 10:33 PM   #5
Jon_Keats
Senior Member
 
Location: Phoenix, AZ

Join Date: Mar 2010
Posts: 279
Default

Thanks for the prompt response.

I don't understand how you get 49
Code:
    Read 1------------------------->
Fragment --------------------------------------------------
                          <--------------------------------Read2
I would call it -49?

On a separate note I just finished our first test of subjunc. I'd say it performed well but had a couple of issues. First I have to say I really like the actual MAPQ value compared to Tophat or STAR. But it did fail on some picard validation steps due to the encoding of the sam/bam file
Code:
Ignoring SAM validation error: ERROR: Record 6, Read name HWI-ST388-W7D:30:D25DKACXX:6:2112:4784:17492, Mate Alignment start should be 0 because reference name = *.
Ignoring SAM validation error: ERROR: Record 6, Read name HWI-ST388-W7D:30:D25DKACXX:6:2112:4784:17492, Mapped mate should have mate reference nameq

HWI-ST388-W7D:30:D25DKACXX:6:2112:4784:17492    129     1       11683   199     100M    *       114359203       0       GAGATTCTTATTAGTGATTTGGGCTGGGGCCTGGCCATGTGTATTTTTTTAAATTTCCACTGATGATTTTGCTGCATGGCCGGTGTTGAGAATGACTGCG    <=?DDADDHDFHHABFHIGHE>ECG<GC@G1??FHICGF<D9BAFEIJJE=F@CEHGIBCHHHEHHHFFFFFEDEEECADCBBB?28<:CDDD@>A:C<<    PG:Z:MarkDuplicate
We typically do markduplicates for our RNAseq runs as a quality control metric but you get similar validation errors until you turn on VALIDATION_STRINGENCY=LENIENT.

Also it looks like the "correctly-paired" flagstats are encoded in a hard-coded variable. In this case these mRNAseq libraries have fragments of 150bp with a SD of 36bp but a read of 146bp insert was called incorrectly paired. I'm not sure if this is the mindist option but this is a basic issue.

On a good note the alignments look decent and for mapping rate it was higher than Tophat2.0.8 and just behind STAR2.3.0

Last edited by Jon_Keats; 06-03-2013 at 10:35 PM.
Jon_Keats is offline   Reply With Quote