Minimal coverage for accurate variant calling according to Illumina FDA submission?

[alunem]

Dear all,
looking up the FDA submission (510K) of the MiSeq Dx platform document, it is stated that the "Recommended minimal coverage per amplicon needed for accurate variant calling (Q(max_gt | poly_site) >=100) is 75x:



My questions are
1 - Is there a known rational for this 75x number?
2 - What does "Q(max_gt | poly_site) >=100" mean?

Thanks!

[Brian Bushnell]

If you are not getting accurate calls at 30x coverage over a particular site, 75x won't help you much, because there's some fundamental problem with your methodology. And I think by 75x per amplicon they mean 150x for a human, which seems quite excessive. I imagine that number is designed so that you can screw up a lot of stuff (reduce library complexity by over-amplifying too little input DNA, use exome-capture and an inferior aligner to yield extreme reference-bias, etc) and still have some chance of producing useful data.

[alunem]

Thanks Brian,
my focus is more on the regulatory point of view though (this is why I was interested by the FDA submission document) so I still need some references/pointers.
Do you understand this?

Q(max_gt | poly_site)

[Brian Bushnell]

Sorry, no, I have not the slightest idea what that means.

[GenoMax]

Thanks Brian,
my focus is more on the regulatory point of view though (this is why I was interested by the FDA submission document) so I still need some references/pointers.
Do you understand this?

http://www.illumina.com/documents/pr...provements.pdf

See page 7 for explanation of Q(max_gt | poly_site) .

[alunem]

Oh great, thanks.