DeepRT2 — RT Prediction Model

A proteomics-based retention time (RT) prediction model built on top of ESM protein language models.

Environment Setup

pip install fair-esm

ESM Model Weights

Download the pre-trained ESM model weights from the official repository:

https://github.com/facebookresearch/esm?tab=readme-ov-file#pre-trained-models-

RT Prediction Data

Data After autoRT Processing

https://github.com/bzhanglab/AutoRT/tree/master/example/data

Raw autoRT Data

source from https://www.ebi.ac.uk/pride/archive/projects/PXD006109. Specifically, the article uses evidence.txt from this directory.

Model Data Preprocessing

• Script: data_preprocess.py
• Note: For convenience, many absolute paths are hard-coded in the script. You will need to modify these paths to match your local environment.

Model Pre-training

• Script: run_model_training.py
• Note: For convenience, many absolute paths are hard-coded in the script. You will need to modify these paths to match your local environment.