Fig. 2From: Transition-transversion encoding and genetic relationship metric in ReliefF feature selection improves pathway enrichment in GWASReliefF combinations of metrics and diffs. We use three methods (two new) for computing the attribute diff value (Eqs. 4, 5, 8): GM, AM, and TiTv (genotype mismatch, allele mismatch, and transition/transversion) with the flag --snp-metric-diff in our inbix software. These three diffs also can be combined with the Manhattan metric (Eq. 9) to create three methods for computing the nearest neighbor distances in the space of all SNPs with the --snp-metric-nn flag combined with --manhattan (Euclidean is also an option but we focus on Manhattan). The fourth nearest neighbor method is the genetic relationship matrix, which does not use the diffs and is called by the grm option with --snp-metric-nn flag. We focus on six combinations of nearest neighbor and diff methods: each of the three diffs with Manhattan for nearest neighbors and each of the three diffs used with GRM to compute nearest neighborsBack to article page