Table 1.
Summary of the results of splice prediction programs for detecting the wild-type splice sites, and the effects of variants on each
| Was the variant predicted to alter the splice site by | NTRK1 c.575–19 G > A Exon 6 acceptor site | NTRK1 c.717 + 4 A > T Exon 6 donor site | SCN9A c.377 + 5 C > T Exon 3 donor site |
|---|---|---|---|
| MaxEntScan (1–16) | Yes: 5 to 3.7 | Yes: 7.2 to 1.7 | No: ND |
| SpliceSiteFinder-like (0–100) | No: 75 to 75 | No: 81 to 71 | No: ND |
| Human Splicing Finder (0–100) | No: 82.5 to 82.5 | No: 91 to 82 | No: ND |
| GeneSplicer (0–15) | No: 2.5 to 2.5 | Yes: 9.2 to 2.5 | No: ND |
| NNSplice (0–1) | No: 0.9 to 0.9 | Yes: 0.7 to 0 | No: ND |
| Spliceman (0–100) | Yes: 75 | Yes: 55 | Yes: 82 |
The Alamut program, which incorporates MaxEntScan, SpliceSiteFinder-like, Human Splicing Finder, GeneSplicer and NNSplice, was used as the primary assessment tool for the intronic variants. For each prediction program in Alamut we have indicated the scale range in brackets and have stated whether the variants were predicted to be deleterious in bold if the change in variation was greater than 15% (yes/no) and given the score of the wild-type splice site and the effect of the variant below the prediction. At least three of the five programs had to strongly predict an effect on splicing at the canonical splice site in order to be considered as disease-causing by the clinical laboratory.
The NTRK1 c.575–19 G > A variant was only predicted to have deleterious effects on the wildtype acceptor site by one program, MaxEntScan. The NTRK1 c.717 + 4 A > T variant had deleterious predictions in three programs but as the +4 position was known to be the least conserved it was not flagged as being likely to affect function. The SCN9A exon 3 splice site was not detected as a splice site by any of the five programs in Alamut, and hence the effect of the variant could not be determined. This was unsurprising as the intron 3 of SCN9A is a U12 intron, for which none of the Alamut programs were designed. However, the more recently written splice prediction program Spliceman did predicted the splice site and the deleterious effect of the variant. Spliceman reports variants as a percentage; the higher the percentile rank, the more likely it is the variation will disrupt splicing.
ND: wildtype splice was not detected by the program