dms_data

Experimental data

Each dataset has its own directory containing the following:

• The dataset .tsv file, a tab-separated dataframe with columns for the variant, number of mutations, and functional score.
• A splits directory containing train, validation, and test splits that have been saved as files for reuse and reproducibility. See the notebook train_test_split.ipynb for an example of how to create these splits.

Additionally, there is a dataset index file, datasets.yml, with additional information about each dataset, such as the name, directory, and wild-type sequence.

Using your own dataset

To use your own dataset, you will need to format it using our standard format and add an entry for it to datasets.yml. I recommend placing raw data in the data/raw_dms_data directory and adding a function to the parse_raw_dms_data.py script to process the dataset into the required format, but this is not strictly necessary (just good practice).

You will need to create train, validation, and test splits for your dataset. See the notebook train_test_split.ipynb for an example of how to do this.

Note
The variants in experimental datasets are expected to be 0-indexed, meaning a mutation in the first position of the sequence should be listed as X0X, not X1X. Future versions of this repository may move to a 1-based indexing system, which is more common in for protein variants.