Unsampled intraspecific variation as a potential source of the missing variation is also unlikely.
Finally, the missing variation is attributed to unique environmental effects plus error.
In any case, these three unsampled taxa certainly cannot account for all the missing variation, i.e., the second 'unknown Western' and the 'unknown Eastern' ribotype, and at least three divergent hybrid chloroplast haplotypes (even if one of them corresponds to an 'unknown' ETS lineage).
This suggests that even if only 70% of the variations were correctly retained by SNPs&GO, the missed variations are probably the less discriminative between germline variations associated to cancer and other diseases.
Had the coverage threshold been 10×, we would have identified seven of these missed variations, but we would have also included seven more miscalls, increasing the number of false positives (Supplementary Table 3 [see Additional file 2]).
There could also be missed variation due to incomplete coverage of genes or due to our limited number of SNPs.
While our approach here may have missed variation of low abundance transcripts and small magnitude expression differences, we identified numerous candidate genes for further study.
Because the measurements from DNA controls reflect the complete process of allele-specific read assessment, the error can originate from properties of the genome (ambiguity, missed variation), mapping, or sequencing related technical bias.
