View Single Post
Old 05-31-2013, 10:42 AM   #81
Jon_Keats
Senior Member
 
Location: Phoenix, AZ

Join Date: Mar 2010
Posts: 279
Default

Somatic Variant Calling

This is a bit of a shameless plug for one of our institutional tools developed for somatic variant calling. In our experience it is one of the best tools for paired tumor-normal/constitutional identification of somatic variants.

Check it out:
https://sites.google.com/site/seuratsomatic/

On other notes, we have been doing a decent bit of testing comparing Tophat and STAR for RNA alignments. We should be posting more in the future but here are my take aways
1) Using updated genomes and GTF dramatically improves the TOPHAT alignment rates (we moved to the new 1000G reference hs37d5 from the inital version and from ensembl 64 to ensembl 70)
2) STAR is smoking fast
3) STAR does a much better job of aligning indels, particularly large indels (this might be a tophat settings issue, it doesn't seem that we really leverage the bowtie2 gapped alignments in tophat... I could be wrong)
4) Expression estimates of same samples aligned with STAR or TOPHAT are nearly identical (r2>0.97)
Jon_Keats is offline   Reply With Quote