Contamination check

Every statement about variants, HLA, pharmacogenetics or haplogroups assumes the sample comes from one person. An unnoticed mixture creates apparent heterozygosity and false findings. Even a few percent of foreign DNA can flip genotypes and produce false positive signals. That is why the contamination check belongs at the start of the analysis and not at its end.

Genome runs VerifyBamID2 on the aligned BAM or CRAM file together with the matching reference genome. The method is genotype-free: it does not require a prior known genotype of the person, but models the observed allele patterns at known variant positions against population allele frequencies. Ancestry is co-estimated from a PCA reference, so the contamination estimate stays largely ancestry-agnostic and holds even when the contaminating DNA comes from a different population. As its metric the tool reports the FREEMIX value, the estimated fraction of foreign DNA, which Genome reads from the result file and surfaces in the report.

In short-read sequencing one does not see a whole foreign person, only slightly raised allele rates across many positions that are hard to separate from sequencing errors and mapping artefacts. Low contamination and low read depth look similar, so a reliable value depends on having enough well-covered, informative positions. VerifyBamID2 addresses this by jointly evaluating many markers with known frequencies and carrying ancestry inside the model rather than assuming it. Genome presents this technical estimate and deliberately keeps it separate from a diagnosis: an elevated value is a warning flag for the interpretation of all further reports, not a finding in itself.

VerifyBamID2 measures human-to-human mixture, that is foreign DNA from another person, but not non-human or microbial reads. Genome covers this second, different quality question elsewhere: in the optional viral analysis, Kraken2 assigns reads taxonomically and Bracken estimates abundance at species level, which makes off-target and non-human signals visible. The two axes complement each other but are separate steps in Genome with their own purpose: the contamination check itself remains solely the task of VerifyBamID2.