Skip to main content
. 2019 Oct 15;20:496. doi: 10.1186/s12859-019-3026-8

Table 4.

Top variant statistics. This table shows the ranking performance statistics for all methods evaluated on our test set (same order as Table 3)

Ranking System Percentage in Top X Variants - X=(1, 10, 20)
All (n=189) VUS (n=111) LP (n=42) Path. (n=36)
CADD Scaled 4, 17, 24 0, 9, 15 7, 21, 30 13, 41, 47
HPO-cosine 7, 32, 47 7, 31, 48 7, 28, 40 8, 38, 50
Exomiser(hiPhive) 7, 29, 36 6, 30, 36 2, 16, 28 16, 38, 44
Exomiser(hiPhive, human only) 7, 28, 37 6, 28, 36 2, 16, 30 16, 38, 50
Phen-Gen 4, 21, 30 5, 20, 27 4, 16, 26 2, 27, 44
DeepPVP 11, 42, 52 4, 36, 47 16, 42, 50 27, 61, 72
RandomForest(sklearn) 16, 53, 65 9, 45, 55 19, 61, 76 36, 69, 80
LogisticRegression(sklearn) 23, 58, 72 13, 44, 62 26, 71, 80 52, 88, 94
BalancedRandomForest(imblearn) 16, 55, 67 9, 44, 57 23, 66, 76 33, 77, 86
EasyEnsembleClassifier(imblearn) 17, 58, 70 12, 43, 60 14, 71, 78 36, 88, 94

The “Percentage in Top X Variants” columns show the percentage of reported variants that were found in the top 1, 10, and 20 variants in a case after ranking by the corresponding method