Tool to identify within-species contamination

Two tools identified:

1. Haplocheck.
   • It estimates contamination by detecting polymorphic sites in the mtDNA data and classifies them into mitochondrial haplogroups.
   • It can be used as a proxy tool to estimate the nDNA contamination levels.
2. VerifyBamID2
   • A method that accurately estimates DNA contamination and is agnostic to genetic ancestry of the intended or contaminating sample.
   • Integrates the estimation of genetic ancestry and DNA contamination in a unified likelihood framework by leveraging individual-specific allele frequencies projected from reference genotypes onto principal component coordinates.

For more on this topic, also see the issue here.

Which tool to use?

Both haplocheck and VerifyBamID2 are currently integrated into QuaC, and they were run across all existing WGS samples.

Here are the samples that these tools identified as contaminated:

Most of them belonged to a particular project. All the ones that were identified by VerifyBamID2 were also flagged by haplocheck, but the reverse is not true.

Looked into that project's samples to see if there was a trend.

In the above image, yellow highlighted samples are Italy samples and the non-highlighted ones are UAB's. Those in bold are probands. In summary, all Italy samples are affected by contamination. While both tools identified them, quantified level of contamination doesn't match. Three UAB samples were also flagged by haplocheck, and only one of them is proband. While it is not clear why only haplocheck has flagged these UAB samples, it is important to note that haplocheck is a proxy tool to estimate the nDNA contamination levels, unlike VerifyBamID2. It may be the case that only MT was affected by heteroplasmy and not contamination, but I haven't done any further digging in this direction.

After a brief discussion with Brandon, it was decided that we will use VerifyBamID2 for contamintation detection. On the topic of using haplocheck just to identify mitochondrial haplogroups, it was decided to use a tool that is meant for such haplohgroup identification. We will be dropping haplocheck from our toolkit (Haplocheck will continue to live in our hearts as it helped kick-off inclusion of this QC in our pipeline 😆).

closed

mentioned in commit c202de84