Home

Maury was run on a set of 1055 tumor-normal pairs. Read depths at 126 common polymorphisms were used to ascertain genotype concordance between the pairs. The SNPS used in the analysis were obtained from the Pengelly et al paper - http://www.ncbi.nlm.nih.gov/pubmed/24070238

The tumor-normal pair which was a known case of sample-swap, and which motivated this package, clearly stands out in the histogram with a score of 0.375.

A couple of the tumor-normal pairs, I think, were actually mom-child pairs from a trio, these are around 0.6(these pairs were created as an exercise to evaluate somatic-mutation calling). The tumors frequently have CNV events and structural-variants which would explain some of the discordance(<1) observed. As you can see the majority of the tumor-normal pairs have perfect concordance at these sites.

A further improvement would be to obtain a probabilistic score instead of a naive concordance metric. One way to do this would be using Likelihood ratio tests and Fisher's method to aggregate the p-values.