Skip to main content
Fig. 7 | BioData Mining

Fig. 7

From: Transition-transversion encoding and genetic relationship metric in ReliefF feature selection improves pathway enrichment in GWAS

Fig. 7

Pathway detection comparison of TiTv-TiTv ReliefF and GRM-TiTv ReliefF. Bars count the number of genes that overlap the given pathway – Axon Guidance (light blue), Neuronal System (dark blue), and G protein-coupled receptor (GPCR) (green) – from the top 500 genes from each feature selection method. The methods are TiTv-TiTv (ReliefF with TiTv-based nearest neighbors (Eqs. 8 and 9) and TiTv attribute diff (Eq. 8)), GRM-TiTv (ReliefF with GRM-based nearest neighbors (Eq. 10) and TITv attribute diff (Eq. 8)), where TiTv is transition/transversion diff and GRM is genetic relationship metric. In addition to ReliefF methods, we compare with Lasso with principal component covariates, Random Forest, and Random Genes (random sampling of 500 genes averaged over 100 replicates). TiTv-TiTv has a similar pattern to but better performance than Random Forest. GRM-TiTv finds the most GPCR genes, and TiTv-TiTv finds the most Axon and Neuronal genes

Back to article page

BioData Mining

ISSN: 1756-0381