> ## Documentation Index
> Fetch the complete documentation index at: https://proto.evodesign.org/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# SpliceTransformer intron boundary score
> Score intron-boundary prediction with SpliceTransformer on three segments concatenated into one 1-kb target.
**License:** SpliceTransformer is open source and free for academic and commercial use under an Apache-2.0 license. Please refer to [the license](https://github.com/ShenLab-Genomics/SpliceTransformer/blob/main/LICENSE) for full terms.
This constraint is open source. Any third-party models, product names, or trademarks referenced are the property of their respective owners, and Proto is not affiliated with them.
```bibtex theme={null}
@article{you2024splicetransformer,
title={SpliceTransformer predicts tissue-specific splicing linked to human diseases},
author={You, Ningyuan and Liu, Chang and Gu, Yuxin and Wang, Rong and Jia, Hanying and Zhang, Tianyun and Jiang, Song and Shi, Jinsong and Chen, Ming and Guan, Min-Xin and Sun, Siqi and Pei, Shanshan and Liu, Zhihong and Shen, Ning},
journal={Nature Communications},
volume={15},
number={1},
pages={9129},
year={2024},
doi={10.1038/s41467-024-53088-6}
}
```
Score donor/acceptor splice sites for three-segment intron boundaries.
Accepts three segments (left\_flank, intron\_core, right\_flank), concatenates
them into a single 1-kb target sequence, and scores donor/acceptor splice
sites.
## API Reference
Configuration for SpliceTransformer intron boundary constraint.
This class defines configuration parameters for evaluating splice site quality
using SpliceTransformer, a deep learning model trained to predict splice sites
in pre-mRNA sequences. The constraint assesses whether specified positions
in a sequence are likely to function as authentic splice sites, which is critical
for proper intron removal and mRNA processing.
SpliceTransformer requires sequences of specific lengths:
* Concatenated target (left\_flank + intron\_core + right\_flank): scored
over a 1000 bp window. If the concatenation is longer, it is windowed
to 1000 bp centred on the donor/acceptor positions and the positions
are remapped into that window.
* Left context: Must be exactly 4000 bp
* Right context: Must be exactly 4000 bp
* Total sequence analyzed: 9000 bp (4000 + 1000 + 4000)
The model outputs splice site probabilities for each position. Higher
scores indicate stronger predicted splice sites. Authentic splice sites
typically have scores > 0.5, while non-splice positions have scores near 0.
Sequence of the left context for SpliceTransformer
Sequence of the right context for SpliceTransformer
0-indexed position(s) into the concatenated target sequence of expected donor
0-indexed position(s) into the concatenated target sequence of expected acceptor
Combine donor/acceptor probs: 'mean' (average) or 'min' (weaker site must be strong).
Options: `mean`, `min`
Max donor/acceptor probability within +/- this many positions of the index; 0 = exact index.
Advanced parameter configuration for SpliceTransformer.
ReturnsConstraintOutput
One result per input. `score` is the combined
boundary penalty in `[0.0, 1.0]`. `metadata` carries `donor_pos`,
`acceptor_pos`, `donor_score`, `acceptor_score`, and
`total_splice_score`.
evo-design/proto-language/proto\_language/constraint/rna\_splicing/splice\_transformer\_intron\_boundary.py
