DNA Sample tracking

[rjohnp]

Hi all,

a not particularly intensive question here. We recently had a couple of sample identities mixed by our exome sequencing provider (i.e. we got back all our data but some of the .fastq files were misidentified). This is obviously a huge issue.

I was wondering what sort of approaches were taken in other core labs in order to prevent, or at least catch this sort of error? If we're supposedly moving towards clinical sequencing as a standard this really needs not to happen.

Thanks for any suggestions.

[Bukowski]

I like the fact you assume it was your sequencing provider that mixed up the sequences, and not your lab that mixed up the samples

I'm only joking, because I'm sure you've gone back and genotyped what you've sent and had your provider genotype the aliquots you sent them but I've had to deal with many situations where we have been sent samples (or data) that is not what the originator claims, at any point in the process where a human is involved there exists the possibility for sample mix up.

I think the only answer from the lab side is automation. You could have a genotyping panel run on samples as they come in, and checked against e.g. exome results on the way out. It would be great to have some kind of sample barcoding that is applied before capture and library prep is carried out that can be assayed later (I'm a bioinformatician, not running the machines, so no idea how feasible this is).

Far too often sample mix ups only present on analysis. My favourite, as it is the easiest to pick up and rectify is the trio's where a parent has been swapped for a child.

[rjohnp]

A valid point of course, but yes, we did check that all.

Very interesting actually, there are some good articles out there on lab automation, mainly for clinical path labs, some showing that error rates actually go up following automation, at least for a couple of months. The hardware is only as good as the programmer and the user, garbage in, garbage out and all that.

I guess an issue with ligating an oligo tag onto all your fragmented DNA (I think that's what you're suggesting here) is that you're essentially reducing you're coverage as you will just trim that sequence of on analysis, so say a 6% loss in coverage for a hexamer tag on 100 bp reads. And if you mix up the samples doing that you have no hope...