Setting twice the background error rate as upper bound of the binomial confidence interval, only 0.0041% and 0.00002% of the variants are expected to be miscalled as true variant on the Illumina MiSeq and Ion Torrent PGM, respectively.
However, if only one of the two reactions fail in a two reaction/ single allele per well format, a heterozygote could be miscalled as a homozygote.
Genotype calling uncertainty in GBS typically arises from sequencing error, which generates miscalls in SNP alleles, and low sequencing depth, which causes heterozygotes to be miscalled as homozygotes.
The pyrosequencing chemistry adopted by Roche GS FLX Titanium sequencing has a weak point with homopolymer sequences (repeats of a single base) that tend to be miscalled.
Assuming that a mismatch occurs when and only when one of the two SNPs in the pair is miscalled, the estimated error rate (E) is obtained from the proportion of SNP pairs with mismatched genotypes (M = 2[ E - E]), which gives E = 1 − 1 − 2 M 2 ≈ M 2 for small mismatch rates.
"And considering that by our statistics, there is only one missed call every five games, three challenges a game ought to more than cover game-changing dramatic plays that are miscalled".
For example SNPs typed at 1 position across two strands, hets are miscalled 2.8% and homozygotes each at 0.4%.
We also measure the positive predictive value (PPV), or the fraction of masked bases that actually were miscalled, and hence, were correctly masked.
$$\end{documentt} Pr { D j | A } = 1 − ζ j D j = A ζ j Pr { A is true | D j is miscalled } otherwise, where Pr{A is true| D j is miscalled} is a precomputed, sequencer specific lookup table.
Such analyses can identify regions that were assembled incorrectly, or even point mutations that were miscalled in one strain.
