| license: mit | |
| configs: | |
| - config_name: dataset | |
| data_files: "dataset.csv" | |
| # Description | |
| Stability Landscape Prediction is a regression task where each input protein *x* is mapped to a label *y* ∈ *R* measuring the most extreme circumstances in which protein *x* maintains its fold above a concentration threshold (a proxy for intrinsic stability). | |
| **Protein Format:** AA sequence | |
| # Splits | |
| The dataset is from [**Evaluating Protein Transfer Learning with TAPE**](https://arxiv.org/abs/1906.08230). We follow the original data splits, with the number of training, validation and test set shown below: | |
| - Train: 53614 | |
| - Valid: 2512 | |
| - Test: 12851 | |
| # Label | |
| The label is the fitness of the protein measuring the most extreme circumstances in which the protein maintains its fold above a concentration threshold (a proxy for intrinsic stability). |