Skip to main content
Fig. 6 | Genome Biology

Fig. 6

From: Benchmarking splice variant prediction algorithms using massively parallel splicing assays

Fig. 6

Influence of masking and annotation choice. (A) Venn charts showing counts of background set variants (n=500,000) called as SDV only when masking is disabled (red), enabled (blue), or in both cases (overlap, purple) for SpliceAI and Pangolin. (B) Same for Pangolin, run with masking using two different annotation sets. C Tracks of SpliceAI scores (y-axis) for all SNVs in FGFR2 exon IIIc showing either score using masking and default annotation (upper panel) or scores with masking and only the FGFR2c isoform (lower panel) vs hg38 position (x-axis), formatted as in Fig. 2. Symbols denote known benign or pathogenic variants from ClinVar or published reports. A cluster of pathogenic exon IIIc acceptor disrupting variants is missed when annotation does not include exon IIIc (yellow region), and exon IIIc donor disrupting variants have intermediate scores (blue region). D SpliceAI masked scores perfectly separate known pathogenic and benign variants when exon IIIc is included but not using default annotations

Back to article page