Open sbilobram opened 5 years ago
Hi!
Thanks for your interest in Pisces.
I don't have any hard coded limits in Pisces. The true min depth you need would be a function of the minimum frequency you need to detect, the confidence you need, and a sophisticated error model tuned to your particular experiment. The minDepth of 10 is quite low, with the expectation that whomever is designing the assay will make sure they have enough depth over their target regions.
As a rough minimum, I would say 30x - 100x for germline, and 1000x+ for somatic. Rule of thumb: If you are looking for a variant at a min frequency of Y=1/Z, make sure your coverage depth allows you to statistically confidently expect to see that variant at least a few times in the reads (~3Z). Ie, think about how many lottery tickets (reads) you need to buy, to be likely to win when your chance with each single ticket is Y. (Then consider that a gremlin is non-randomly pilfering lottery tickets, and replacing them with junk - basecalling errors, alignment errors, etc...)
best Tamsen
Okay great information. And I can say I got this info from a Pisces developer?
yes.
Another question ...different topic : Does Pisces give a VAF for the SNVs?
Nevermind found these as VF in the INFO fileld
We potentially would like to try Pisces in a project with low-coverage BAMs. I do not find mention of input limits and only that minDepth default is at 10 and adjustable via parameter -c . Is there a recommended Coverage for Pisces inputs (BAMs) ? Minimum coverage?