The scores range from 0 to 1 and can be interpreted as the probability of the variant being splice-altering. In the paper, a detailed characterization is provided for 0.2 (high recall), 0.5 (recommended), and 0.8 (high precision) cutoffs.
++These tracks are in bigWig format. The signal height represents the SpliceAI probability score. +This track may be configured in a variety of ways to highlight different aspects of the displayed +information. Click the "Graph configuration help" link for an explanation of configuration +options.
+According to the strength of their supporting evidence, variants were classified as "splice-altering" (~25%), "not splice-altering" (~25%), and "low-frequency splice-altering" (~50%), which correspond to weak or indeterminate evidence of spliceogenicity. 55% of the splice-altering variants in SpliceVarDB are outside the canonical splice sites (5.6% are deep intronic). The data is shown as lollipop plots that can be clicked, the details page then shows a link to SpliceVarDB with full details.
The classification thresholds primarily follow those established by the original study. However, most studies only defined criteria for splice-altering variants and did not define criteria for variants that resulted in normal splicing. The authors implemented stringent thresholds to define the normal category and ensure a high-quality set of control variants. Variants that did not meet these criteria were classified as low-frequency splice-altering