Support

Posted by jgSoton on Mon 09 of July, 2012 18:54 MYT

novoalign version differences?

Posted by jgSoton on Mon 09 of July, 2012 18:54 MYT

Hi,

I've been using novoalign novoalignMPI (V2.07.11 - Build May 27 2011 @ 15:31:23 - A short read aligner with qualities.

And have recently updated to novoalignMPI (V2.08.01 - Build Mar 14 2012 @ 10:25:47 - A short read aligner with qualities.

I've noticed this seems to have had quite an effect on the number of variants called further down my pipeline (I'm using Varscan for variant calling).

Some example output:
One lane of data aligned...
v2.08.01 v2.07.11

Paired Reads 14889536 14889536
Pairs Aligned 12053765 12026592
Read Sequences 29779072 29779072
Aligned 28196793 28144000
Unique Alignment 25858796 25812347
Gapped Alignment 371157 242721
Quality Filter 263431 277622
Homopolymer Filter 1841 1978

Varscan output files...
v2.08.01 v2.07.11
SOTON0015_SOTON0016.snp 71159 71090
SOTON0015_SOTON0016.indel 6681 2520
SOTON0015_SOTON0016_filtered 69567 69652
SOTON0015_SOTON0016_filtered.Germline 67515 67567
SOTON0015_SOTON0016_filtered.LOH 580 577
SOTON0015_SOTON0016_filtered.Somatic 1401 1437
SOTON0015_SOTON0016_filtered.Somatic.hc 287 294

The mapping stats seem broadly similar, but I get many more indels called and due to a filtering step in Varscan (which removes SNPs in indels) I seem to get less high confidence variants at the end.

As updating Novoalign is the only difference between these 2 runs I wonder if you have an explanation?

Thanks for any help.
Jane

Posted by colin on Mon 09 of July, 2012 22:26 MYT

Posted by colin on Mon 09 of July, 2012 22:26 MYT

Hi Jane,

At one point we reduced the default gap extend penalty from 15 to 6. Thisresults in a few more gapped alignments. You can set it back to the previous behavious with -x 15.

This should be documented in the readme.txt file.

Kind Regards, Colin