Athurva Gore/LabNotes/2009-8-28
Jump to navigation
Jump to search
[edit]
To Do[edit]
- Forms for Tech Transfer Office
- Invoice for Zymo
Removing Reads with Multiple SNPS[edit]
DF699[edit]
- Took results from removal of reads w/SNPs in the last 15 bp
- Removed reads that had more than one SNP.
- 9386 hits, 1103 misses
- Don't do any better than just removing those with SNPs in the last 15 bp.
- Probably not the approach to take.
Remaining options[edit]
- Considered:
- Removing reads w/multiple SNPs
- Removing reads w/SNPs near the end (removes many bad SNPs, but also some good SNPs)
- Trimming reads w/SNPs near the end (removes many bad SNPs as well; retains more good and bad though)
- Still remaining:
- Reducing quality score of SNPs - probably won't see any better results than above
- New Idea: Split reads into two sections; first 25, last 15?
- For now, we will go with removing reads with SNPs near the end
IPS and Cancer[edit]
- For Proposal, preliminary data is needed of the difference between DF699 and fibroblast.
- Want this data to be high quality
- Started with results that removed reads w/ SNPs withing 15 bp of the end
- Then ran pileup2variantsPair with min coverage of 8 and min quality of 20.
- Used a chi-squared test and removed SNPs that scored lower than 6.
- Found:
- 193 SNPs
- 145 not at dbSNP locations (the above 48 are probably just miscalls in one case due to biased capture)
- 128 Loss-of-Heterozygosity
- 13 Gain-of-Heterozygosity
- 4 Doubles
- If these GOH mutations are investigated, only 5 are nonsynonymous
- 3 of these appear in the COSMIC set
- Decided to check both LOH and GOH mutations to see if any are valid.
- SIFT: Does it report ref/call? Or does it report in same order?
- None are classified as purely damaging by SIFT; one is damaging but with low confidence
- Need to investigate data itself, check over SNPs.
- Decided to compare to a filter where SNPs had a minimum coverage of 4 instead
- Will pass just GOH to SIFT, check to see if analysis for presentation was correct