Fig 1.
Selection of genes, using the COSMIC Cancer Gene Census—Haematopoietic and Lymphoid Tissue and EpiFactors database.
There were at least 50 AAS in dbSNP and COSMIC for each of these 19 selected genes (in the red circle) that further constituted our dataset.
Fig 2.
EpiMut procedure that was applied to each of the 19 proteins in the dataset.
Table 1.
The variants dataset consisted of 2881 variants in 19 epigenetic factors.
Fig 3.
Comparison between the variant effect predictor based on Gene Specific Models (GSM) and the Multiple Genes Model (MGM).
Fig 4.
Performance of EpiMut, PolyPhen-2, SIFT and SNAP2 on a dataset consisting of variants in epigenetic factors mutated in hematologic malignancies.
Fig 5.
EpiMut is contrasted to SIFT, PolyPhen-2 and SNAP2 in the search for unshared correctly classified mutations.
Stacked bars represent the numbers of exclusive TPs in each of the three comparisons.
Fig 6.
Distribution per gene of 162 “difficult to predict mutations” (DTP) predicted by EpiMut, PolyPhen-2, SIFT and SNAP2.
Table 2.
nCFD dataset consisting of 2108 variants in non-conserved regions of epigenetic factors.
Fig 7.
Differences in performance of EpiMut, PolyPhen-2, SIFT and SNAP2 between the entire variants dataset and nCFD data subset.
Each column in this histogram represents the difference among the values of a particular performance measure obtained, for each tool, on the entire variants dataset and the values obtained for the nCFD data subset.