| --- |
| library_name: multimolecule |
| license: agpl-3.0 |
| pipeline: mean-ribosome-load |
| pipeline_tag: other |
| tags: |
| - Biology |
| - RNA |
| - 5' UTR |
| - Translation |
| - rna |
| widget: |
| - example_title: microRNA 21 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: UAGCUUAUCAGACUGAUGUUGA |
| - example_title: microRNA 146a |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: UGAGAACUGAAUUCCAUGGGUU |
| - example_title: microRNA 155 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: UUAAUGCUAAUCGUGAUAGGGGUU |
| - example_title: RNA component of mitochondrial RNA processing endoribonuclease |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: GGUUCGUGCUGAAGGCCUGUAUCCUAGGCUACACACUGAGGACUCUGUUCCUCCCCUUUCCGCCUAGGGGAAAGUCCCCGGACCUCGGGCAGAGAGUGCCACGUGCAUACGCACGUAGACAUUCCCCGCUUCCCACUCCAAAGUCCGCCAAGAAGCGUAUCCCGCUGAGCGGCGUGGCGCGGGGGCGUCAUCCGUCAGCUCCCUCUAGUUACGCAGGCAGUGCGUGUCCGCGCACCAACCACACGGGGCUCAUUCUCAGCGCGGCUGUAAAAAAAAA |
| - example_title: 7SK small nuclear RNA |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: GGAUGUGAGGGCGAUCUGGCUGCGACAUCUGUCACCCCAUUGAUCGCCAGGGUUGAUUCGGCUGAUCUGGCUGGCUAGGCGGGUGUCCCCUUCCUCCCUCACCGCUCCAUGUGCGUCCCUCCCGAAGCUGCGCGCUCGGUCGAAGAGGACGACCAUCCCCGAUAGAGGAGGACCGGUCUUCGGUCAAGGGUAUACGAGUAGCUGCGCUCCCCUGCUAGAACCUCCAAACAAGCUCUCAAGGUCCAUUUGUAGGAGAACGUAGGGUAGUCAAGCUUCCAAGACUCCAGACACAUCCAAAUGAGGCGCUGCAUGUGGCAGUCUGCCUUUCUUUU |
| - example_title: telomerase RNA component |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: GGGUUGCGGAGGGUGGGCCUGGGAGGGGUGGUGGCCAUUUUUUGUCUAACCCUAACUGAGAAGGGCGUAGGCGCCGUGCUUUUGCUCCCCGCGCGCUGUUUUUCUCGCUGACUUUCAGCGGGCGGAAAAGCCUCGGCCUGCCGCCUUCCACCGUUCAUUCUAGAGCAAACAAAAAAUGUCAGCUGCUGGCCCGUUCGCCCCUCCCGGGGACCUGCGGCGGGUCGCCUGCCCAGCCCCCGAACCCCGCCUGGAGGCCGCGGUCGGCCCGGGGCUUCUCCGGAGGCACCCACUGCCACCGCGAAGAGUUGGGCUCUGUCAGCCGCGGGUCUCUCGGGGGCGAGGGCGAGGUUCAGGCCUUUCAGGCCGCAGGAAGAGGAACGGAGCGAGUCCCCGCGCGCGGCGCGAUUCCCUGAGCUGUGGGACGUGCACCCAGGACUCGGCUCACACAUGC |
| - example_title: vault RNA 2-1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: CGGGUCGGAGUUAGCUCAAGCGGUUACCUCCUCAUGCCGGACUUUCUAUCUGUCCAUCUCUGUGCUGGGGUUCGAGACCCGCGGGUGCUUACUGACCCUUUUAUGCAA |
| - example_title: brain cytoplasmic RNA 1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: GGCCGGGCGCGGUGGCUCACGCCUGUAAUCCCAGCUCUCAGGGAGGCUAAGAGGCGGGAGGAUAGCUUGAGCCCAGGAGUUCGAGACCUGCCUGGGCAAUAUAGCGAGACCCCGUUCUCCAGAAAAAGGAAAAAAAAAAACAAAAGACAAAAAAAAAAUAAGCGUAACUUCCCUCAAAGCAACAACCCCCCCCCCCCUUU |
| - example_title: HIV-1 TAR-WT |
| pipeline_tag: mean-ribosome-load |
| sequence_type: ncRNA |
| task: mean-ribosome-load |
| text: GGUCUCUCUGGUUAGACCAGAUCUGAGCCUGGGAGCUCUCUGGCUAACUAGGGAACC |
| - example_title: prion protein (Kanno blood group) |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AUGGCGAACCUUGGCUGCUGGAUGCUGGUUCUCUUUGUGGCCACAUGGAGUGACCUGGGCCUCUGC |
| - example_title: interleukin 10 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AUGCACAGCUCAGCACUGCUCUGUUGCCUGGUCCUCCUGACUGGGGUGAGGGCC |
| - example_title: Zaire ebolavirus |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AAUGUUCAAACACUUUGUGAAGCUCUGUUAGCUGAUGGUCUUGCUAAAGCAUUUCCUAGCAAUAUGAUGGUAGUCACAGAGCGUGAGCAAAAAGAAAGCUUAUUGCAUCAAGCAUCAUGGCACCACACAAGUGAUGAUUUUGGUGAGCAUGCCACAGUUAGAGGGAGUAGCUUUGUAACUGAUUUAGAGAAAUACAAUCUUGCAUUUAGAUAUGAGUUUACAGCACCUUUUAUAGAAUAUUGUAACCGUUGCUAUGGUGUUAAGAAUGUUUUUAAUUGGAUGCAUUAUACAAUCCCACAGUGUUAU |
| - example_title: SARS coronavirus |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AUGUUUAUUUUCUUAUUAUUUCUUACUCUCACUAGUGGUAGUGACCUUGACCGGUGCACCACUUUUGAUGAUGUUCAAGCUCCUAAUUACACUCAACAUACUUCAUCUAUGAGGGGGGUUUACUAUCCUGAUGAAAUUUUUAGAUCAGACACUCUUUAUUUAACUCAGGAUUUAUUUCUUCCAUUUUAUUCUAAUGUUACAGGGUUUCAUACUAUUAAUCAUACGUUUGACAACCCUGUCAUACCUUUUAAGGAUGGUAUUUAUUUUGCUGCCACAGAGAAAUCAAAUGUUGUCCGUGGUUGGGUUUUUGGUUCUACCAUGAACAACAAGUCACAGUCGGUGAUUAUUAUUAACAAUUCUACUAAUGUUGUUAUACGAGCAUGUAACUUUGAAUUGUGUGACAACCCUUUCUUUGCUGUUUCUAAACCCAUGGGUACACAGACACAUACUAUGAUAUUCGAUAAUGCAUUUAAAUGCACUUUCGAGUACAUAUCU |
| - example_title: insulin |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AUGGCCCUGUGGAUGCGCCUCCUGCCCCUGCUGGCGCUGCUGGCCCUCUGGGGACCUGACCCAGCCGCAGCCUUUGUGAACCAACACCUGUGCGGCUCACACCUGGUGGAAGCUCUCUACCUAGUGUGCGGGGAACGAGGCUUCUUCUACACACCCAAGACCCGCCGGGAGGCAGAGGACCUGCAGGUGGGGCAGGUGGAGCUGGGCGGGGGCCCUGGUGCAGGCAGCCUGCAGCCCUUGGCCCUGGAGGGGUCCCUGCAGAAGCGUGGCAUUGUGGAACAAUGCUGUACCAGCAUCUGCUCCCUCUACCAGCUGGAGAACUACUGCAACUAG |
| - example_title: cyclin dependent kinase inhibitor 2A |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AUGGAGCCGGCGGCGGGGAGCAGCAUGGAGCCUUCGGCUGACUGGCUGGCCACGGCCGCGGCCCGGGGUCGGGUAGAGGAGGUGCGGGCGCUGCUGGAGGCGGGGGCGCUGCCCAACGCACCGAAUAGUUACGGUCGGAGGCCGAUCCAGGUCAUGAUGAUGGGCAGCGCCCGAGUGGCGGAGCUGCUGCUGCUCCACGGCGCGGAGCCCAACUGCGCCGACCCCGCCACUCUCACCCGACCCGUGCACGACGCUGCCCGGGAGGGCUUCCUGGACACGCUGGUGGUGCUGCACCGGGCCGGGGCGCGGCUGGACGUGCGCGAUGCCUGGGGCCGUCUGCCCGUGGACCUGGCUGAGGAGCUGGGCCAUCGCGAUGUCGCACGGUACCUGCGCGCGGCUGCGGGGGGCACCAGAGGCAGUAACCAUGCCCGCAUAGAUGCCGCGGAAGGUCCCUCAGACAUCCCCGAUUGA |
| - example_title: human papillomavirus type 16 E6 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: mRNA |
| task: mean-ribosome-load |
| text: AUGCACCAAAAGAGAACUGCAAUGUUUCAGGACCCACAGGAGCGACCCAGAAAGUUACCACAGUUAUGCACAGAGCUGCAAACAACUAUACAUGAUAUAAUAUUAGAAUGUGUGUACUGCAAGCAACAGUUACUGCGACGUGAGGUAUAUGACUUUGCUUUUCGGGAUUUAUGCAUAGUAUAUAGAGAUGGGAAUCCAUAUGCUGUAUGUGAUAAAUGUUUAAAGUUUUAUUCUAAAAUUAGUGAGUAUAGACAUUAUUGUUAUAGUUUGUAUGGAACAACAUUAGAACAGCAAUACAACAAACCGUUGUGUGAUUUGUUAAUUAGGUGUAUUAACUGUCAAAAGCCACUGUGUCCUGAAGAAAAGCAAAGACAUCUGGACAAAAAGCAAAGAUUCCAUAAUAUAAGGGGUCGGUGGACCGGUCGAUGUAUGUCUUGUUGCAGAUCAUCAAGAACACGUAGAGAAACCCAGCUGUAA |
| - example_title: NRAS proto-oncogene |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 5' UTR |
| task: mean-ribosome-load |
| text: GGGGCCGGAAGUGCCGCUCCUUGGUGGGGGCUGUUCAUGGCGGUUCCGGGGUCUCCAACAUUUUUCCCGGCUGUGGUCCUAAAUCUGUCCAAAGCAGAGGCAGUGGAGCUUGAGGUUCUUGCUGGUGUGAA |
| - example_title: amyloid beta precursor protein |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 5' UTR |
| task: mean-ribosome-load |
| text: GUCAGUUUCCUCGGCAGCGGUAGGCGAGAGCACGCGGAGGAGCGUGCGCGGGGGCCCCGGGAGACGGCGGCGGUGGCGGCGCGGGCAGAGCAAGGACGCGGCGGAUCCCACUCGCACAGCAGCGCACUCGGUGCCCCGCGCAGGGUCGCG |
| - example_title: RUNX family transcription factor 1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 5' UTR |
| task: mean-ribosome-load |
| text: ACUUCUUUGGGCCUCAUAAACAACCACAGAACCACAAGUUGGGUAGCCUGGCAGUGUCAGAAGUCUGAACCCAGCAUAGUGGUCAGCAGGCAGGACGAAUCACACUGAAUGCAAACCACAGGGUUUCGCAGCGUGGUAAAAGAAAUCAUUGAGUCCCCCGCCUUCAGAAGAGGGUGCAUUUUCAGGAGGAAGCG |
| - example_title: fragile X messenger ribonucleoprotein 1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 5' UTR |
| task: mean-ribosome-load |
| text: CUCAGUCAGGCGCUCAGCUCCGUUUCGGUUUCACUUCCGGUGGAGGGCCGCCUCUGAGCGGGCGGCGGGCCGACGGCGAGCGCGGGCGGCGGCGGUGACGGAGGCGCCGCUGCCAGGGGGCGUGCGGCAGCGCGGCGGCGGCGGCGGCGGCGGCGGCGGCGGAGGCGGCGGCGGCGGCGGCGGCGGCGGCGGCUGGGCCUCGAGCGCCCGCAGCCCACCUCUCGGGGGCGGGCUCCCGGCGCUAGCAGGGCUGAAGAGAAG |
| - example_title: MYC proto-oncogene |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 5' UTR |
| task: mean-ribosome-load |
| text: AACUCGCUGUAGUAAUUCCAGCGAGAGGCAGAGGGAGCGAGCGGGCGGCCGGCUAGGGUGGAAGAGCCGGGCGAGCAGAGCUGCGCUGCGGGCGUCCUGGGAAGGGAGAUCCGGAGCGAAUAGGGGGCUUCGCCUCUGGCCCAGCCCUCCCGCUGAUCCCCCAGCCAGCGGUCCGCAACCCUUGCCGCAUCCACGAAACUUUGCCCAUAGCAGCGGGCGGGCACUUUGCACUGGAACUUACAACACCCGAGCAAGGACGCGACUCUCCCGACGCGGGGAGGCUAUUCUGCCCAUUUGGGGACACUUCCCCGCCGCUGCCAGGACCCGCUUCUCUGAAAGGCUCUCCUUGCAGCUGCUUAGACG |
| - example_title: activating transcription factor 4 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 5' UTR |
| task: mean-ribosome-load |
| text: CAUUUCUACUUUGCCCGCCCACAGAUGUAGUUUUCUCUGCGCGUGUGCGUUUUCCCUCCUCCCCGCCCUCAGGGUCCACGGCCACCAUGGCGUAUUAGGGGCAGCAGUGCCUGCGGCAGCAUUGGCCUUUGCAGCGGCGGCAGCAGCACCAGGCUCUGCAGCGGCAACCCCCAGCGGCUUAAGCCAUGGCGCUUCUCACGGCAUUCAGCAGCAGCGUUGCUGUAACCGACAAAGACACCUUCGAAUUAAGCACAUUCCUCGAUUCCAGCAAAGCACCGCAAC |
| - example_title: Human GPI protein p137 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 3' UTR |
| task: mean-ribosome-load |
| text: UUUUUAAAAGGAAAAGAUACCAAAUGCCUGCUGCUACCACCCUUUUCAAUUGCUAUGUUUUGAAAGGCACCAGUAUGUGUUUUAGAUUGAUUUAAAUGUUUCAUUUAAAUCACGGACAGUAGUUUCAGUUCUGAUGGUAUAAGCAAAACAAAUAAAACGUUUAUAAAAGUUGUAUCUUGAAACACUGGUGUUCAACAGCUAGCAGCUUAUGUGAUUCACCCCAUGCCACGUUAGUGUCACAAAUUUUAUGGUUUAUCUCCAGCAACAUUUCUCUAGUACUUGCACUUAUUAUCUGAAUUC |
| - example_title: nucleophosmin 1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 3' UTR |
| task: mean-ribosome-load |
| text: GAAAAUAGUUUAAACAAUUUGUUAAAAAAUUUUCCGUCUUAUUUCAUUUCUGUAACAGUUGAUAUCUGGCUGUCCUUUUUAUAAUGCAGAGUGAGAACUUUCCCUACCGUGUUUGAUAAAUGUUGUCCAGGUUCUAUUGCCAAGAAUGUGUUGUCCAAAAUGCCUGUUUAGUUUUUAAAGAUGGAACUCCACCCUUUGCUUGGUUUUAAGUAUGUAUGGAAUGUUAUGAUAGGACAUAGUAGUAGCGGUGGUCAGACAUGGAAAUGGUGGGGAGACAAAAAUAUACAUGUGAAAUAAAACUCAGUAUUUUAAUAAAGUAGCACGGUUUCUAUUGA |
| - example_title: superoxide dismutase 1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 3' UTR |
| task: mean-ribosome-load |
| text: ACAUUCCCUUGGAUGUAGUCUGAGGCCCCUUAACUCAUCUGUUAUCCUGCUAGCUGUAGAAAUGUAUCCUGAUAAACAUUAAACACUGUAAUCUUAAAAGUGUAAUUGUGUGACUUUUUCAGAGUUGCUUUAAAGUACCUGUAGUGAGAAACUGAUUUAUGAUCACUUGGAAGAUUUGUAUAGUUUUAUAAAACUCAGUUAAAAUGUCUGUUUCAAUGACCUGUAUUUUGCCAGACUUAAAUCACAGAUGGGUAUUAAACUUGUCAGAAUUUCUUUGUCAUUCAAGCCUGUGAAUAAAAACCCUGUAUGGCACUUAUUAUGAGGCUAUUAAAAGAAUCCAAAUUCAAACUAAA |
| - example_title: hemoglobin subunit alpha 2 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 3' UTR |
| task: mean-ribosome-load |
| text: CUGGAGCCUCGGUAGCCGUUCCUCCUGCCCGCUGGGCCUCCCAACGGGCCCUCCUCCCCUCCUUGCACCGGCCCUUCCUGGUCUUUGAAUAAAGUCUGAGUGGGCAGCA |
| - example_title: BRAF proto-oncogene |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 3' UTR |
| task: mean-ribosome-load |
| text: AACAAAUGAGUGAGAGAGUUCAGGAGAGUAGCAACAAAAGGAAAAUAAAUGAACAUAUGUUUGCUUAUAUGUUAAAUUGAAUAAAAUACUCUCUUUUUUUUUAAGGUGAACCAAAGAACACUUGUGUGGUUAAAGACUAGAUAUAAUUUUUCCCCAAACUAAAAUUUAUACUUAACAUUGGAUUUUUAACAUCCAAGGGUUAAAAUACAUAGACAUUGCUAAAAAUUGGCAGAGCCUCUUCUAGAGGCUUUACUUUCUGUUCCGGGUUUGUAUCAUUCACUUGGUUAUUUUAAGUAGUAAACUUCAGUUUCUCAUGCAACUUUUGUUGCCAGCUAUCACAUGUCCACUAGGGACUCCAGAAGAAGACCCUACCUAUGCCUGUGUUUGCAGGUGAGAAGUUGGCAGUCGGUUAGCCUGGG |
| - example_title: H3 clustered histone 1 |
| pipeline_tag: mean-ribosome-load |
| sequence_type: 3' UTR |
| task: mean-ribosome-load |
| text: UUACUGUGGUCUCUCUGACGGUCCAAGCAAAGGCUCUUUUCAGAGCCACCACCUUUUC |
| --- |
| |
| # OptMRL |
|
|
| Convolutional neural network for predicting the mean ribosome load (MRL) of an mRNA from the 50 nucleotides upstream of the coding sequence. |
|
|
| ## Disclaimer |
|
|
| This is an UNOFFICIAL implementation of [Interpreting Deep Neural Networks for the Prediction of Translation Rates](https://doi.org/10.1101/2023.06.02.543405) by Frederick Korbel, et al. |
|
|
| The OFFICIAL repository of OptMRL is at [ohlerlab/mlcis](https://github.com/ohlerlab/mlcis). |
|
|
| > [!TIP] |
| > The MultiMolecule team has confirmed that the provided model and checkpoints are producing the same intermediate representations as the original implementation. |
|
|
| **The team releasing OptMRL did not write this model card for this model so this model card has been written by the MultiMolecule team.** |
|
|
| ## Model Details |
|
|
| OptMRL is a small 1D convolutional neural network trained to predict the mean ribosome load (MRL), a polysome-profiling-derived translation efficiency proxy, from the 50 nucleotides of 5' untranslated region (5'UTR) sequence immediately upstream of the coding sequence. The model was first pre-trained on roughly 260,000 random 5'UTR reporters and then fine-tuned on roughly 20,000 endogenous human 5'UTRs. Please refer to the [Training Details](#training-details) section for more information on the training process. |
|
|
| The architecture is a stack of three `Conv1D` layers (120 filters, kernel size 8, `same` padding, ReLU activation) followed by a `Flatten`, a 40-unit `Dense` bottleneck with ReLU activation and dropout, and a final scalar `Dense` regression head. |
|
|
| ### Model Specification |
|
|
| | Num Layers | Hidden Size | Num Parameters (M) | FLOPs (M) | MACs (M) | Max Num Tokens | |
| | ---------- | ----------- | ------------------ | --------- | -------- | -------------- | |
| | 5 | 40 | 0.476 | 24.04 | 12.00 | 50 | |
|
|
| ### Links |
|
|
| - **Code**: [multimolecule.optmrl](https://github.com/DLS5-Omics/multimolecule/tree/master/multimolecule/models/optmrl) |
| - **Data**: 260,000 random 5'UTR reporters (pre-training) + 20,000 human 5'UTR reporters (fine-tuning) |
| - **Paper**: [Interpreting Deep Neural Networks for the Prediction of Translation Rates](https://doi.org/10.1101/2023.06.02.543405) |
| - **Developed by**: Frederick Korbel, Ekaterina Eroshok, Uwe Ohler |
| - **Model type**: 1D CNN for mean-ribosome-load regression from 5'UTR sequence |
| - **Original Repository**: [ohlerlab/mlcis](https://github.com/ohlerlab/mlcis) |
|
|
| ## Usage |
|
|
| The model file depends on the [`multimolecule`](https://multimolecule.danling.org) library. You can install it using pip: |
|
|
| ```bash |
| pip install multimolecule |
| ``` |
|
|
| ### Direct Use |
|
|
| #### Mean Ribosome Load Prediction |
|
|
| You can use this model directly to predict the mean ribosome load of a 50-nucleotide 5'UTR window: |
|
|
| ```python |
| >>> from multimolecule import RnaTokenizer, OptMrlForSequencePrediction |
| |
| >>> tokenizer = RnaTokenizer.from_pretrained("multimolecule/optmrl") |
| >>> model = OptMrlForSequencePrediction.from_pretrained("multimolecule/optmrl") |
| >>> sequence = "ACGU" * 12 + "AC" # 50 nt |
| >>> input = tokenizer(sequence, add_special_tokens=False, return_tensors="pt") |
| >>> output = model(**input) |
| |
| >>> output.logits.shape |
| torch.Size([1, 1]) |
| ``` |
|
|
| ### Interface |
|
|
| - **Input length**: 50 nt fixed 5'UTR window taken immediately upstream of the coding sequence |
| - **Padding**: shorter sequences are right-padded with zeros to 50 nt; longer sequences are truncated to the first 50 nt |
| - **Alphabet**: `ACGU` only; unknown / `N` tokens contribute zero one-hot signal |
| - **Special tokens**: do not add (`add_special_tokens=False`) |
| - **Output**: single scalar mean-ribosome-load (MRL) score per window |
|
|
| ## Training Details |
|
|
| OptMRL was first pre-trained on a large random-5'UTR reporter library and then fine-tuned on a smaller library of endogenous human 5'UTRs. |
|
|
| ### Training Data |
|
|
| - **Pre-training**: ~260,000 random 5'UTR reporters paired with polysome-profiling MRL measurements. |
| - **Fine-tuning**: ~20,000 endogenous human 5'UTR reporters paired with polysome-profiling MRL measurements. |
|
|
| Each reporter contributes a 50-nucleotide 5'UTR window immediately upstream of the coding sequence and a scalar MRL label. |
|
|
| Note [`RnaTokenizer`][multimolecule.RnaTokenizer] will convert "T"s to "U"s for you, you may disable this behaviour by passing `replace_T_with_U=False`. |
|
|
| ### Training Procedure |
|
|
| #### Pre-training |
|
|
| The model was first pre-trained as a regression task to predict the measured MRL of each random 5'UTR reporter, then fine-tuned end-to-end on the human-5'UTR reporters using the same regression objective. The published model is the fine-tuned model. |
|
|
| ## Citation |
|
|
| ```bibtex |
| @article{korbel2023interpreting, |
| author = {Korbel, Frederick and Eroshok, Ekaterina and Ohler, Uwe}, |
| title = {Interpreting Deep Neural Networks for the Prediction of Translation Rates}, |
| journal = {bioRxiv}, |
| publisher = {Cold Spring Harbor Laboratory}, |
| year = {2023}, |
| doi = {10.1101/2023.06.02.543405} |
| } |
| ``` |
|
|
| > [!NOTE] |
| > The artifacts distributed in this repository are part of the MultiMolecule project. |
| > If MultiMolecule supports your research, please cite the MultiMolecule project as follows: |
|
|
| ```bibtex |
| @software{chen_2024_12638419, |
| author = {Chen, Zhiyuan and Zhu, Sophia Y.}, |
| title = {MultiMolecule}, |
| doi = {10.5281/zenodo.12638419}, |
| publisher = {Zenodo}, |
| url = {https://doi.org/10.5281/zenodo.12638419}, |
| year = 2024, |
| month = may, |
| day = 4 |
| } |
| ``` |
|
|
| ## Contact |
|
|
| Please use GitHub issues of [MultiMolecule](https://github.com/DLS5-Omics/multimolecule/issues) for any questions or comments on the model card. |
|
|
| Please contact the authors of the [OptMRL paper](https://doi.org/10.1101/2023.06.02.543405) for questions or comments on the paper/model. |
|
|
| ## License |
|
|
| This model implementation is licensed under the [GNU Affero General Public License](license.md). |
|
|
| For additional terms and clarifications, please refer to our [License FAQ](license-faq.md). |
|
|
| ```spdx |
| SPDX-License-Identifier: AGPL-3.0-or-later |
| ``` |
