Fine-Tuning Protein Language Models with Deep Mutational Scanning Improves Variant Effect Prediction
Abstract
Protein Language Models (PLMs) have emerged as performant and scalable tools for predicting the functional impact and clinical significance of protein-coding variants, but they still lag experimental accuracy. Here, we present a novel finetuning approach to improve the performance of PLMs with experimental maps of variant effects from Deep Mutational Scanning (DMS) assays using a Normalised Log-odds Ratio (NLR) head. We find consistent improvements in a held-out protein test set, and on independent DMS and clinical variant annotation benchmarks from ProteinGym and ClinVar. These findings demonstrate that DMS is a promising source of sequence diversity and supervised training data for improving the performance of PLMs for variant effect prediction.
Cite
Text
Lafita et al. "Fine-Tuning Protein Language Models with Deep Mutational Scanning Improves Variant Effect Prediction." ICLR 2024 Workshops: MLGenX, 2024.Markdown
[Lafita et al. "Fine-Tuning Protein Language Models with Deep Mutational Scanning Improves Variant Effect Prediction." ICLR 2024 Workshops: MLGenX, 2024.](https://mlanthology.org/iclrw/2024/lafita2024iclrw-finetuning/)BibTeX
@inproceedings{lafita2024iclrw-finetuning,
title = {{Fine-Tuning Protein Language Models with Deep Mutational Scanning Improves Variant Effect Prediction}},
author = {Lafita, Aleix and Gonzalez, Ferran and Hossam, Mahmoud and Smyth, Paul and Deasy, Jacob and Allyn-Feuer, Ari L. and Seaton, Daniel D and Young, Stephen},
booktitle = {ICLR 2024 Workshops: MLGenX},
year = {2024},
url = {https://mlanthology.org/iclrw/2024/lafita2024iclrw-finetuning/}
}