TABLE 5.

Analysis of the repeated features over different runs. The column headers display each of the features that appear in Table 4. The rows contain information on whether these features were present or not in each of the performed runs: XX indicates that the feature was repeated across common sets of features, and X indicates the presence of the feature in only one common set.

	KRFPC435	KRFPC566	KRFPC3948	KRFPC4830	KRFPC413	KRFPC608	KRFPC3399	KRFPC3899
Initial run (Table 3)	-	X	-	X	X	X	-	-
Extra five runs (Supplementary Table S2)	-	XX	XX	X	-	XX	-	-
Change in cross validation split (Supplementary Table S5)	XX	XX	XX	X	-	-	-	X
Change in cross validation split (Supplemenatry Table S6)	-	X	-	-	-	XX	-	-
Change in cross validation split (Supplementary Table S7)	XX	-	-	X	-	X	-	-
Change initial population (Supplemenatary Table S8)	X	XX	XX	-	-	-	X	XX
Change initial population (Supplementary Table S9)	XX	XX	X	X	X	-	-	-
Change initial population (Supplementary Table S10)	X	XX	-	X	-	-	-	-
Feature importance	0.53	0.8	0.47	0.4	0.13	0.4	0.07	0.2