Results
Closed this issue · 2 comments
Hi, I've run deTIN on a truth-set of artificially created TIN from 1-30%, for two samples. The results file is attached. Two thoughts;
- I seem to get higher score with aSCGC than with SSNV.
- aSCG at very low levels of containation seem to report artificially high values.
Have you got any advice to offer to improve/interpret the results?
Also, I've noticed that even with 1% TIN contamination my SNV calling looses a lot of accuracy, do you believe deTIN can be tuned to discover such low levels of TIN?
Many thanks for your input!
I set the tin-prior to 0.01 for all samples, to try to pull them towards the lower solution
Hey Magalena,
Looks good overall though I agree that 1 and 2 look could be issues. I think it might be easier for me to offer advice via a google hangout or longer emails. Can you drop me an email at amaro@broadinstitute.org so we can discuss?