Fig. 2

Number of important features (SNPs) identified across the eight training subsets of variable sample sizes from RFECV. Each column represents the different training subsets, and each row represents the individual features. Features are row-sorted based on the number of subsets that they were commonly identified in, with each block separated by a pale-blue divider (i.e., the first block of features, highlighted by a red box, represented the SNPs that were identified across all eight subsets based on the RFECV algorithm). Intensity of red represent the importance of the feature (based on the feature importance score) within each subset; Black represents features that were not identified to be important in the respective subset