Sample marking for exon capture sequencing

[lijr]

Hi every one,

We sent 100 cases and 100 controls samples to the sequencing company for exon sequencing data, we are afraid that the company might make mistakes when they sent the data back. If one sample is belong to the case group and the company misidentified it and sent us the data as a control sample, we will get wrong results. So we want to know if there are some methods to mark a sample before building library and we can identify each sample from the sequecing data?

Thanks.

[Heisman]

If they are doing the entire prep, then no.

However, post-analysis you can look for variants that only occur in a single individual. Then you can Sanger sequence for that variant in the specific individual. That would confirm it is the right sample. This would be very annoying with 200 people as you would pick 200 different samples, but it would work. However, if this company has a decent reputation, then it is probable that they will not make any mistakes, and if they do and realize it they will let you know.

[swbarnes2]

The easy thing to check; see if the SNPs on Chr X and Chr Y agree with the sexes of the samples. If they all match, there were probably no mix-ups.

[lijr]

Thank you all to answer my question.

I just wonder which kind of variants should we choose and how many SNPs should we choose.

Moreover, we also have samples in one family. Usually the case and control in one family are brothers and sisters. The members in one family may have some common variants. The SNPs on ChrX and ChrY can figure out ther samples of different gender. But if two brothers' sample are mixed, how to figure that out?

I ask this question for the company has mixed up our samples. Fortunately, they mixed up the samples of the mother and her son, so we can figure it out by the SNPs on ChrY. We are afraid that they may make the mistake again in future.

Thanks again for all your reply.

[Heisman]

Thank you all to answer my question.

I just wonder which kind of variants should we choose and how many SNPs should we choose.

Moreover, we also have samples in one family. Usually the case and control in one family are brothers and sisters. The members in one family may have some common variants. The SNPs on ChrX and ChrY can figure out ther samples of different gender. But if two brothers' sample are mixed, how to figure that out?

I ask this question for the company has mixed up our samples. Fortunately, they mixed up the samples of the mother and her son, so we can figure it out by the SNPs on ChrY. We are afraid that they may make the mistake again in future.

Thanks again for all your reply.

With whole exome data, you'll find tons of variants that are heterozygous in only one of the parents and heterozygous in only one of the children. Then, if you still have source DNA, you can sanger sequence 1 of those specific positions and determine which data set that person belongs too.

[swbarnes2]

I don't think there's any way, other than genotypng the original DNA for SNPs that will uniquely identify the sample.

I guess you could try to spike in some foreign DNA into the human sample, so that you'd see that in your Illumina data, but I don't know if anyone actaully adulterates their human DNA samples like that.

[lijr]

With whole exome data, you'll find tons of variants that are heterozygous in only one of the parents and heterozygous in only one of the children. Then, if you still have source DNA, you can sanger sequence 1 of those specific positions and determine which data set that person belongs too.

Thanks for your reply.

It is a good method to identify each sample. However, it need primers for each person. I wonder if there are some regions that have lots of polymorphism and can figure everyone out like the barcode. Then we can figure everyone out by a group of primers. Is that available?

Thank you.

[Heisman]

Hmm, if you are confident that the variation is biological and not due to aligning errors, then that could work, but I'm skeptical you'll find a lot of regions that are really dense with a lot of snps in various individuals. That said, there will be lots of opportunities in an entire exome, so it might be doable.

[lijr]

I think the genes for immune like HLA or some gene like that may have lots of SNPs in population. But I don't know if there are some mature methods for this problem.

Or could I do as swbarnes2 said, spike in some artificial sequences that are not homology with the human sequence samples? The artificial sequences can be synthesized and differ from each other. Then when we analyse the result, we can find the definite SNPs from the definite sample. I just wonder if this will hurt the sample and make some unknown influence on analysis.

Thank you.