Skip to main content

Table 1 Accuracy on SNPs that can be represented with unique kmers vs. SNPs that cannot. The accuracy increases for variants that cannot be represented with unique kmers after implementing idea 1 (KAGE with model of kmer counts) and increases further when introducing idea 2 (KAGE full). The accuracy on SNPs that can be represented with unique kmers is always high, since these SNPs are easy to genotype

From: KAGE: fast alignment-free graph-based genotyping of SNPs and short indels

Method

Accuracy on SNPs with non-unique kmers

Accuracy on SNPs with unique kmers

Naive KAGE

64 %

97 %

KAGE with model of kmer counts

77 %

97 %

KAGE (full)

92 %

99 %