more thresholding questions
18 views
Skip to first unread message
richardC
Dec 21, 2011, 11:53:03 AM12/21/11
to JointSNVMix User Group
Hi again.
I am processing tumour-normal genome pairs.
My intended workflow was to use joint_snv_mix_two, threshold, and
joint_fisher to generate a large list of candidate somatic mutations
and then use mutationSeq to generate some specificity.
Originally, I was just thresholding using the 3 algorithms with
(column 10 + columns 11) > 0.5. However, this create a list of
hundreds of thousands of putative somatic snps. I soon learned that
mutation seq cannot handle more than a few thousand putative somatic
snps, so now I need to find a way to dial back my candidate list.
If I threshold oint_snv_mix_two, and joint_fisher results greater than
0.8 I get just a few thousand results, however, I wonder if I'm
missing anything interesting.
Does anyone out there have good recommendations for paired genome
calling with joinSNVMix to then pass the results to mutatuionSeq?
I am processing tumour-normal genome pairs.
My intended workflow was to use joint_snv_mix_two, threshold, and
joint_fisher to generate a large list of candidate somatic mutations
and then use mutationSeq to generate some specificity.
Originally, I was just thresholding using the 3 algorithms with
(column 10 + columns 11) > 0.5. However, this create a list of
hundreds of thousands of putative somatic snps. I soon learned that
mutation seq cannot handle more than a few thousand putative somatic
snps, so now I need to find a way to dial back my candidate list.
If I threshold oint_snv_mix_two, and joint_fisher results greater than
0.8 I get just a few thousand results, however, I wonder if I'm
missing anything interesting.
Does anyone out there have good recommendations for paired genome
calling with joinSNVMix to then pass the results to mutatuionSeq?
Reply all
Reply to author
Forward
0 new messages