multimolecule/mmsplice

1---2library_name: multimolecule3license: agpl-3.04pipeline: splice-variant-effect5pipeline_tag: other6tags:7- Biology8- RNA9- Splicing10- rna11widget:12- example_title: microRNA 2113  pipeline_tag: splice-variant-effect14  sequence_type: ncRNA15  task: splice-variant-effect16  text: UAGCUUAUCAGACUGAUGUUGA17- example_title: microRNA 146a18  pipeline_tag: splice-variant-effect19  sequence_type: ncRNA20  task: splice-variant-effect21  text: UGAGAACUGAAUUCCAUGGGUU22- example_title: microRNA 15523  pipeline_tag: splice-variant-effect24  sequence_type: ncRNA25  task: splice-variant-effect26  text: UUAAUGCUAAUCGUGAUAGGGGUU27- example_title: RNA component of mitochondrial RNA processing endoribonuclease28  pipeline_tag: splice-variant-effect29  sequence_type: ncRNA30  task: splice-variant-effect31  text: GGUUCGUGCUGAAGGCCUGUAUCCUAGGCUACACACUGAGGACUCUGUUCCUCCCCUUUCCGCCUAGGGGAAAGUCCCCGGACCUCGGGCAGAGAGUGCCACGUGCAUACGCACGUAGACAUUCCCCGCUUCCCACUCCAAAGUCCGCCAAGAAGCGUAUCCCGCUGAGCGGCGUGGCGCGGGGGCGUCAUCCGUCAGCUCCCUCUAGUUACGCAGGCAGUGCGUGUCCGCGCACCAACCACACGGGGCUCAUUCUCAGCGCGGCUGUAAAAAAAAA32- example_title: 7SK small nuclear RNA33  pipeline_tag: splice-variant-effect34  sequence_type: ncRNA35  task: splice-variant-effect36  text: GGAUGUGAGGGCGAUCUGGCUGCGACAUCUGUCACCCCAUUGAUCGCCAGGGUUGAUUCGGCUGAUCUGGCUGGCUAGGCGGGUGUCCCCUUCCUCCCUCACCGCUCCAUGUGCGUCCCUCCCGAAGCUGCGCGCUCGGUCGAAGAGGACGACCAUCCCCGAUAGAGGAGGACCGGUCUUCGGUCAAGGGUAUACGAGUAGCUGCGCUCCCCUGCUAGAACCUCCAAACAAGCUCUCAAGGUCCAUUUGUAGGAGAACGUAGGGUAGUCAAGCUUCCAAGACUCCAGACACAUCCAAAUGAGGCGCUGCAUGUGGCAGUCUGCCUUUCUUUU37- example_title: telomerase RNA component38  pipeline_tag: splice-variant-effect39  sequence_type: ncRNA40  task: splice-variant-effect41  text: GGGUUGCGGAGGGUGGGCCUGGGAGGGGUGGUGGCCAUUUUUUGUCUAACCCUAACUGAGAAGGGCGUAGGCGCCGUGCUUUUGCUCCCCGCGCGCUGUUUUUCUCGCUGACUUUCAGCGGGCGGAAAAGCCUCGGCCUGCCGCCUUCCACCGUUCAUUCUAGAGCAAACAAAAAAUGUCAGCUGCUGGCCCGUUCGCCCCUCCCGGGGACCUGCGGCGGGUCGCCUGCCCAGCCCCCGAACCCCGCCUGGAGGCCGCGGUCGGCCCGGGGCUUCUCCGGAGGCACCCACUGCCACCGCGAAGAGUUGGGCUCUGUCAGCCGCGGGUCUCUCGGGGGCGAGGGCGAGGUUCAGGCCUUUCAGGCCGCAGGAAGAGGAACGGAGCGAGUCCCCGCGCGCGGCGCGAUUCCCUGAGCUGUGGGACGUGCACCCAGGACUCGGCUCACACAUGC42- example_title: vault RNA 2-143  pipeline_tag: splice-variant-effect44  sequence_type: ncRNA45  task: splice-variant-effect46  text: CGGGUCGGAGUUAGCUCAAGCGGUUACCUCCUCAUGCCGGACUUUCUAUCUGUCCAUCUCUGUGCUGGGGUUCGAGACCCGCGGGUGCUUACUGACCCUUUUAUGCAA47- example_title: brain cytoplasmic RNA 148  pipeline_tag: splice-variant-effect49  sequence_type: ncRNA50  task: splice-variant-effect51  text: GGCCGGGCGCGGUGGCUCACGCCUGUAAUCCCAGCUCUCAGGGAGGCUAAGAGGCGGGAGGAUAGCUUGAGCCCAGGAGUUCGAGACCUGCCUGGGCAAUAUAGCGAGACCCCGUUCUCCAGAAAAAGGAAAAAAAAAAACAAAAGACAAAAAAAAAAUAAGCGUAACUUCCCUCAAAGCAACAACCCCCCCCCCCCUUU52- example_title: HIV-1 TAR-WT53  pipeline_tag: splice-variant-effect54  sequence_type: ncRNA55  task: splice-variant-effect56  text: GGUCUCUCUGGUUAGACCAGAUCUGAGCCUGGGAGCUCUCUGGCUAACUAGGGAACC57- example_title: prion protein (Kanno blood group)58  pipeline_tag: splice-variant-effect59  sequence_type: mRNA60  task: splice-variant-effect61  text: AUGGCGAACCUUGGCUGCUGGAUGCUGGUUCUCUUUGUGGCCACAUGGAGUGACCUGGGCCUCUGC62- example_title: interleukin 1063  pipeline_tag: splice-variant-effect64  sequence_type: mRNA65  task: splice-variant-effect66  text: AUGCACAGCUCAGCACUGCUCUGUUGCCUGGUCCUCCUGACUGGGGUGAGGGCC67- example_title: Zaire ebolavirus68  pipeline_tag: splice-variant-effect69  sequence_type: mRNA70  task: splice-variant-effect71  text: AAUGUUCAAACACUUUGUGAAGCUCUGUUAGCUGAUGGUCUUGCUAAAGCAUUUCCUAGCAAUAUGAUGGUAGUCACAGAGCGUGAGCAAAAAGAAAGCUUAUUGCAUCAAGCAUCAUGGCACCACACAAGUGAUGAUUUUGGUGAGCAUGCCACAGUUAGAGGGAGUAGCUUUGUAACUGAUUUAGAGAAAUACAAUCUUGCAUUUAGAUAUGAGUUUACAGCACCUUUUAUAGAAUAUUGUAACCGUUGCUAUGGUGUUAAGAAUGUUUUUAAUUGGAUGCAUUAUACAAUCCCACAGUGUUAU72- example_title: SARS coronavirus73  pipeline_tag: splice-variant-effect74  sequence_type: mRNA75  task: splice-variant-effect76  text: AUGUUUAUUUUCUUAUUAUUUCUUACUCUCACUAGUGGUAGUGACCUUGACCGGUGCACCACUUUUGAUGAUGUUCAAGCUCCUAAUUACACUCAACAUACUUCAUCUAUGAGGGGGGUUUACUAUCCUGAUGAAAUUUUUAGAUCAGACACUCUUUAUUUAACUCAGGAUUUAUUUCUUCCAUUUUAUUCUAAUGUUACAGGGUUUCAUACUAUUAAUCAUACGUUUGACAACCCUGUCAUACCUUUUAAGGAUGGUAUUUAUUUUGCUGCCACAGAGAAAUCAAAUGUUGUCCGUGGUUGGGUUUUUGGUUCUACCAUGAACAACAAGUCACAGUCGGUGAUUAUUAUUAACAAUUCUACUAAUGUUGUUAUACGAGCAUGUAACUUUGAAUUGUGUGACAACCCUUUCUUUGCUGUUUCUAAACCCAUGGGUACACAGACACAUACUAUGAUAUUCGAUAAUGCAUUUAAAUGCACUUUCGAGUACAUAUCU77- example_title: insulin78  pipeline_tag: splice-variant-effect79  sequence_type: mRNA80  task: splice-variant-effect81  text: AUGGCCCUGUGGAUGCGCCUCCUGCCCCUGCUGGCGCUGCUGGCCCUCUGGGGACCUGACCCAGCCGCAGCCUUUGUGAACCAACACCUGUGCGGCUCACACCUGGUGGAAGCUCUCUACCUAGUGUGCGGGGAACGAGGCUUCUUCUACACACCCAAGACCCGCCGGGAGGCAGAGGACCUGCAGGUGGGGCAGGUGGAGCUGGGCGGGGGCCCUGGUGCAGGCAGCCUGCAGCCCUUGGCCCUGGAGGGGUCCCUGCAGAAGCGUGGCAUUGUGGAACAAUGCUGUACCAGCAUCUGCUCCCUCUACCAGCUGGAGAACUACUGCAACUAG82- example_title: cyclin dependent kinase inhibitor 2A83  pipeline_tag: splice-variant-effect84  sequence_type: mRNA85  task: splice-variant-effect86  text: AUGGAGCCGGCGGCGGGGAGCAGCAUGGAGCCUUCGGCUGACUGGCUGGCCACGGCCGCGGCCCGGGGUCGGGUAGAGGAGGUGCGGGCGCUGCUGGAGGCGGGGGCGCUGCCCAACGCACCGAAUAGUUACGGUCGGAGGCCGAUCCAGGUCAUGAUGAUGGGCAGCGCCCGAGUGGCGGAGCUGCUGCUGCUCCACGGCGCGGAGCCCAACUGCGCCGACCCCGCCACUCUCACCCGACCCGUGCACGACGCUGCCCGGGAGGGCUUCCUGGACACGCUGGUGGUGCUGCACCGGGCCGGGGCGCGGCUGGACGUGCGCGAUGCCUGGGGCCGUCUGCCCGUGGACCUGGCUGAGGAGCUGGGCCAUCGCGAUGUCGCACGGUACCUGCGCGCGGCUGCGGGGGGCACCAGAGGCAGUAACCAUGCCCGCAUAGAUGCCGCGGAAGGUCCCUCAGACAUCCCCGAUUGA87- example_title: human papillomavirus type 16 E688  pipeline_tag: splice-variant-effect89  sequence_type: mRNA90  task: splice-variant-effect91  text: AUGCACCAAAAGAGAACUGCAAUGUUUCAGGACCCACAGGAGCGACCCAGAAAGUUACCACAGUUAUGCACAGAGCUGCAAACAACUAUACAUGAUAUAAUAUUAGAAUGUGUGUACUGCAAGCAACAGUUACUGCGACGUGAGGUAUAUGACUUUGCUUUUCGGGAUUUAUGCAUAGUAUAUAGAGAUGGGAAUCCAUAUGCUGUAUGUGAUAAAUGUUUAAAGUUUUAUUCUAAAAUUAGUGAGUAUAGACAUUAUUGUUAUAGUUUGUAUGGAACAACAUUAGAACAGCAAUACAACAAACCGUUGUGUGAUUUGUUAAUUAGGUGUAUUAACUGUCAAAAGCCACUGUGUCCUGAAGAAAAGCAAAGACAUCUGGACAAAAAGCAAAGAUUCCAUAAUAUAAGGGGUCGGUGGACCGGUCGAUGUAUGUCUUGUUGCAGAUCAUCAAGAACACGUAGAGAAACCCAGCUGUAA92- example_title: NRAS proto-oncogene93  pipeline_tag: splice-variant-effect94  sequence_type: 5' UTR95  task: splice-variant-effect96  text: GGGGCCGGAAGUGCCGCUCCUUGGUGGGGGCUGUUCAUGGCGGUUCCGGGGUCUCCAACAUUUUUCCCGGCUGUGGUCCUAAAUCUGUCCAAAGCAGAGGCAGUGGAGCUUGAGGUUCUUGCUGGUGUGAA97- example_title: amyloid beta precursor protein98  pipeline_tag: splice-variant-effect99  sequence_type: 5' UTR100  task: splice-variant-effect101  text: GUCAGUUUCCUCGGCAGCGGUAGGCGAGAGCACGCGGAGGAGCGUGCGCGGGGGCCCCGGGAGACGGCGGCGGUGGCGGCGCGGGCAGAGCAAGGACGCGGCGGAUCCCACUCGCACAGCAGCGCACUCGGUGCCCCGCGCAGGGUCGCG102- example_title: RUNX family transcription factor 1103  pipeline_tag: splice-variant-effect104  sequence_type: 5' UTR105  task: splice-variant-effect106  text: ACUUCUUUGGGCCUCAUAAACAACCACAGAACCACAAGUUGGGUAGCCUGGCAGUGUCAGAAGUCUGAACCCAGCAUAGUGGUCAGCAGGCAGGACGAAUCACACUGAAUGCAAACCACAGGGUUUCGCAGCGUGGUAAAAGAAAUCAUUGAGUCCCCCGCCUUCAGAAGAGGGUGCAUUUUCAGGAGGAAGCG107- example_title: fragile X messenger ribonucleoprotein 1108  pipeline_tag: splice-variant-effect109  sequence_type: 5' UTR110  task: splice-variant-effect111  text: CUCAGUCAGGCGCUCAGCUCCGUUUCGGUUUCACUUCCGGUGGAGGGCCGCCUCUGAGCGGGCGGCGGGCCGACGGCGAGCGCGGGCGGCGGCGGUGACGGAGGCGCCGCUGCCAGGGGGCGUGCGGCAGCGCGGCGGCGGCGGCGGCGGCGGCGGCGGCGGAGGCGGCGGCGGCGGCGGCGGCGGCGGCGGCUGGGCCUCGAGCGCCCGCAGCCCACCUCUCGGGGGCGGGCUCCCGGCGCUAGCAGGGCUGAAGAGAAG112- example_title: MYC proto-oncogene113  pipeline_tag: splice-variant-effect114  sequence_type: 5' UTR115  task: splice-variant-effect116  text: AACUCGCUGUAGUAAUUCCAGCGAGAGGCAGAGGGAGCGAGCGGGCGGCCGGCUAGGGUGGAAGAGCCGGGCGAGCAGAGCUGCGCUGCGGGCGUCCUGGGAAGGGAGAUCCGGAGCGAAUAGGGGGCUUCGCCUCUGGCCCAGCCCUCCCGCUGAUCCCCCAGCCAGCGGUCCGCAACCCUUGCCGCAUCCACGAAACUUUGCCCAUAGCAGCGGGCGGGCACUUUGCACUGGAACUUACAACACCCGAGCAAGGACGCGACUCUCCCGACGCGGGGAGGCUAUUCUGCCCAUUUGGGGACACUUCCCCGCCGCUGCCAGGACCCGCUUCUCUGAAAGGCUCUCCUUGCAGCUGCUUAGACG117- example_title: activating transcription factor 4118  pipeline_tag: splice-variant-effect119  sequence_type: 5' UTR120  task: splice-variant-effect121  text: CAUUUCUACUUUGCCCGCCCACAGAUGUAGUUUUCUCUGCGCGUGUGCGUUUUCCCUCCUCCCCGCCCUCAGGGUCCACGGCCACCAUGGCGUAUUAGGGGCAGCAGUGCCUGCGGCAGCAUUGGCCUUUGCAGCGGCGGCAGCAGCACCAGGCUCUGCAGCGGCAACCCCCAGCGGCUUAAGCCAUGGCGCUUCUCACGGCAUUCAGCAGCAGCGUUGCUGUAACCGACAAAGACACCUUCGAAUUAAGCACAUUCCUCGAUUCCAGCAAAGCACCGCAAC122- example_title: Human GPI protein p137123  pipeline_tag: splice-variant-effect124  sequence_type: 3' UTR125  task: splice-variant-effect126  text: UUUUUAAAAGGAAAAGAUACCAAAUGCCUGCUGCUACCACCCUUUUCAAUUGCUAUGUUUUGAAAGGCACCAGUAUGUGUUUUAGAUUGAUUUAAAUGUUUCAUUUAAAUCACGGACAGUAGUUUCAGUUCUGAUGGUAUAAGCAAAACAAAUAAAACGUUUAUAAAAGUUGUAUCUUGAAACACUGGUGUUCAACAGCUAGCAGCUUAUGUGAUUCACCCCAUGCCACGUUAGUGUCACAAAUUUUAUGGUUUAUCUCCAGCAACAUUUCUCUAGUACUUGCACUUAUUAUCUGAAUUC127- example_title: nucleophosmin 1128  pipeline_tag: splice-variant-effect129  sequence_type: 3' UTR130  task: splice-variant-effect131  text: GAAAAUAGUUUAAACAAUUUGUUAAAAAAUUUUCCGUCUUAUUUCAUUUCUGUAACAGUUGAUAUCUGGCUGUCCUUUUUAUAAUGCAGAGUGAGAACUUUCCCUACCGUGUUUGAUAAAUGUUGUCCAGGUUCUAUUGCCAAGAAUGUGUUGUCCAAAAUGCCUGUUUAGUUUUUAAAGAUGGAACUCCACCCUUUGCUUGGUUUUAAGUAUGUAUGGAAUGUUAUGAUAGGACAUAGUAGUAGCGGUGGUCAGACAUGGAAAUGGUGGGGAGACAAAAAUAUACAUGUGAAAUAAAACUCAGUAUUUUAAUAAAGUAGCACGGUUUCUAUUGA132- example_title: superoxide dismutase 1133  pipeline_tag: splice-variant-effect134  sequence_type: 3' UTR135  task: splice-variant-effect136  text: ACAUUCCCUUGGAUGUAGUCUGAGGCCCCUUAACUCAUCUGUUAUCCUGCUAGCUGUAGAAAUGUAUCCUGAUAAACAUUAAACACUGUAAUCUUAAAAGUGUAAUUGUGUGACUUUUUCAGAGUUGCUUUAAAGUACCUGUAGUGAGAAACUGAUUUAUGAUCACUUGGAAGAUUUGUAUAGUUUUAUAAAACUCAGUUAAAAUGUCUGUUUCAAUGACCUGUAUUUUGCCAGACUUAAAUCACAGAUGGGUAUUAAACUUGUCAGAAUUUCUUUGUCAUUCAAGCCUGUGAAUAAAAACCCUGUAUGGCACUUAUUAUGAGGCUAUUAAAAGAAUCCAAAUUCAAACUAAA137- example_title: hemoglobin subunit alpha 2138  pipeline_tag: splice-variant-effect139  sequence_type: 3' UTR140  task: splice-variant-effect141  text: CUGGAGCCUCGGUAGCCGUUCCUCCUGCCCGCUGGGCCUCCCAACGGGCCCUCCUCCCCUCCUUGCACCGGCCCUUCCUGGUCUUUGAAUAAAGUCUGAGUGGGCAGCA142- example_title: BRAF proto-oncogene143  pipeline_tag: splice-variant-effect144  sequence_type: 3' UTR145  task: splice-variant-effect146  text: AACAAAUGAGUGAGAGAGUUCAGGAGAGUAGCAACAAAAGGAAAAUAAAUGAACAUAUGUUUGCUUAUAUGUUAAAUUGAAUAAAAUACUCUCUUUUUUUUUAAGGUGAACCAAAGAACACUUGUGUGGUUAAAGACUAGAUAUAAUUUUUCCCCAAACUAAAAUUUAUACUUAACAUUGGAUUUUUAACAUCCAAGGGUUAAAAUACAUAGACAUUGCUAAAAAUUGGCAGAGCCUCUUCUAGAGGCUUUACUUUCUGUUCCGGGUUUGUAUCAUUCACUUGGUUAUUUUAAGUAGUAAACUUCAGUUUCUCAUGCAACUUUUGUUGCCAGCUAUCACAUGUCCACUAGGGACUCCAGAAGAAGACCCUACCUAUGCCUGUGUUUGCAGGUGAGAAGUUGGCAGUCGGUUAGCCUGGG147- example_title: H3 clustered histone 1148  pipeline_tag: splice-variant-effect149  sequence_type: 3' UTR150  task: splice-variant-effect151  text: UUACUGUGGUCUCUCUGACGGUCCAAGCAAAGGCUCUUUUCAGAGCCACCACCUUUUC152---153 154# MMSplice155 156Modular modeling of the effects of genetic variants on splicing.157 158## Disclaimer159 160This is an UNOFFICIAL implementation of the [MMSplice: modular modeling improves the predictions of genetic variant effects on splicing](https://doi.org/10.1186/s13059-019-1653-z) by Jun Cheng, et al.161 162The OFFICIAL repository of MMSplice is at [gagneurlab/MMSplice_MTSplice](https://github.com/gagneurlab/MMSplice_MTSplice).163 164> [!TIP]165> The MultiMolecule team has confirmed that the provided model and checkpoints are producing the same intermediate representations as the original implementation.166 167**The team releasing MMSplice did not write this model card for this model so this model card has been written by the MultiMolecule team.**168 169## Model Details170 171MMSplice is a _modular_ neural network for predicting the effect of genetic variants on pre-mRNA splicing. It decomposes an exon together with its flanking introns into five regions and scores each region with an independent small convolutional sub-network. For variant-effect estimation, the model is run on both the reference and the alternative sequence, and the per-module score deltas are combined by a fixed linear model into a delta-logit-PSI splicing-effect score. Please refer to the [Training Details](#training-details) section for more information on the training process.172 173### Model Specification174 175| Num Modules | Num Parameters (M) | FLOPs (M) | MACs (M) |176| ----------- | ------------------ | --------- | -------- |177| 5           | 0.057              | 5.71      | 2.79     |178 179(FLOPs and MACs measured on a 220 bp exon-with-flanks input.)180 181### Links182 183- **Code**: [multimolecule.mmsplice](https://github.com/DLS5-Omics/multimolecule/tree/master/multimolecule/models/mmsplice)184- **Data**: Human splice-site and exon data with MPRA exon-skipping variant-effect measurements185- **Paper**: [MMSplice: modular modeling improves the predictions of genetic variant effects on splicing](https://doi.org/10.1186/s13059-019-1653-z)186- **Developed by**: Jun Cheng, Thi Yen Duong Nguyen, Kamil J. Cygan, Muhammed Hasan Çelik, William G. Fairbrother, Žiga Avsec, Julien Gagneur187- **Model type**: Modular 1D CNN with five region-specific sub-networks for splice variant-effect prediction188- **Original Repository**: [gagneurlab/MMSplice_MTSplice](https://github.com/gagneurlab/MMSplice_MTSplice)189 190## Usage191 192The model file depends on the [`multimolecule`](https://multimolecule.danling.org) library. You can install it using pip:193 194```bash195pip install multimolecule196```197 198### Direct Use199 200#### Module Scores201 202```python203>>> import torch204>>> from multimolecule import RnaTokenizer, MmSpliceForSequencePrediction205 206>>> tokenizer = RnaTokenizer.from_pretrained("multimolecule/mmsplice")207>>> model = MmSpliceForSequencePrediction.from_pretrained("multimolecule/mmsplice")208>>> _ = model.eval()209>>> left_intron = "A" * 100210>>> exon = "C" * 20211>>> right_intron = "G" * 100212>>> reference = tokenizer(left_intron + exon + right_intron, add_special_tokens=False, return_tensors="pt")213>>> output = model.model(**reference)214>>> output["logits"].shape215torch.Size([1, 5])216```217 218#### Variant Effect219 220```python221>>> import torch222>>> from multimolecule import RnaTokenizer, MmSpliceForSequencePrediction223 224>>> tokenizer = RnaTokenizer.from_pretrained("multimolecule/mmsplice")225>>> model = MmSpliceForSequencePrediction.from_pretrained("multimolecule/mmsplice")226>>> _ = model.eval()227>>> left_intron = "A" * 100228>>> exon = "C" * 20229>>> right_intron = "G" * 100230>>> reference = tokenizer(left_intron + exon + right_intron, add_special_tokens=False, return_tensors="pt")231>>> alternative_exon = exon[:10] + "U" + exon[11:]232>>> alternative = tokenizer(left_intron + alternative_exon + right_intron, add_special_tokens=False, return_tensors="pt")233>>> output = model(234...     reference["input_ids"],235...     alternative_input_ids=alternative["input_ids"],236... )237>>> output["logits"].shape238torch.Size([1, 1])239```240 241### Interface242 243- **Input length**: exon sequence with 100 nt upstream intronic context + 100 nt downstream intronic context244- **Tokenization**: disable special tokens; the embedding layer maps `A/C/G/U` ids to the four upstream channels and maps `N`, padding, special, and unknown tokens to all-zero columns245- **Output (reference-only call, `input_ids` / `inputs_embeds`)**: per-module score vector `logits` of shape `(batch_size, 5)`246 247### Variant Effect248 249- **Reference + alternative call** (also pass `alternative_input_ids` / `alternative_inputs_embeds`): additionally returns `alternative_logits` and per-module `delta_logits = alternative_logits - logits`250- **`MmSpliceForSequencePrediction`**: requires both reference and alternative; returns the combined scalar delta-logit-PSI score of shape `(batch_size, 1)`251 252## Training Details253 254MMSplice was trained as five independent modules on splicing data and the modules were combined with a linear model to predict variant effects on percent-spliced-in (PSI).255 256### Training Data257 258The acceptor, donor, exon, and intron modules were trained on splice-site and exon data derived from human reference transcripts. The combining linear model was fit against a massively parallel reporter assay (MPRA) of exon-skipping variants.259 260### Training Procedure261 262#### Pre-training263 264Each module was trained with a sequence-to-scalar objective scoring its region. The module scores (and their reference/alternative deltas) were then combined by a fixed linear model into a delta-logit-PSI splicing-effect score.265 266## Citation267 268```bibtex269@article{cheng2019mmsplice,270  title     = {MMSplice: modular modeling improves the predictions of genetic variant effects on splicing},271  author    = {Cheng, Jun and Nguyen, Thi Yen Duong and Cygan, Kamil J and {\c{C}}elik, Muhammed Hasan and Fairbrother, William G and Avsec, {\v{Z}}iga and Gagneur, Julien},272  journal   = {Genome Biology},273  volume    = 20,274  number    = 1,275  pages     = {48},276  year      = 2019,277  publisher = {Springer},278  doi       = {10.1186/s13059-019-1653-z}279}280```281 282> [!NOTE]283> The artifacts distributed in this repository are part of the MultiMolecule project.284> If MultiMolecule supports your research, please cite the MultiMolecule project as follows:285 286```bibtex287@software{chen_2024_12638419,288  author    = {Chen, Zhiyuan and Zhu, Sophia Y.},289  title     = {MultiMolecule},290  doi       = {10.5281/zenodo.12638419},291  publisher = {Zenodo},292  url       = {https://doi.org/10.5281/zenodo.12638419},293  year      = 2024,294  month     = may,295  day       = 4296}297```298 299## Contact300 301Please use GitHub issues of [MultiMolecule](https://github.com/DLS5-Omics/multimolecule/issues) for any questions or comments on the model card.302 303Please contact the authors of the [MMSplice paper](https://doi.org/10.1186/s13059-019-1653-z) for questions or comments on the paper/model.304 305## License306 307This model implementation is licensed under the [GNU Affero General Public License](license.md).308 309For additional terms and clarifications, please refer to our [License FAQ](license-faq.md).310 311```spdx312SPDX-License-Identifier: AGPL-3.0-or-later313```