ProCeSa

This is the source code of "ProCeSa: Contrast-Enhanced Structure-Aware Network for Thermostability Prediction with Protein Language Models" (https://doi.org/10.1021/acs.jcim.4c01752).

conda environment

/procesa/environment.yml

esm model

Download pretrained ESM1B-650M, ESMC-600M model from original github. Put these models in /dwnl_ckpts.

data generation

1. Download the 'procesa_data.zip' (https://drive.google.com/file/d/1gYO1l_HJugmDxAa1JOdKnuojYqaF15Ur/view?usp=sharing), uncompress it, and put the folder in /dataset. It should look like /dataset/procesa_data/...

2. Use scripts in /procesa/FLIP/baselines/scripts/ to generate dgl graph pkl files. Generated data will be saved in /datasets.

Train and evaluate

Run scripts in /procesa/scripts to train and evaluate models. Results will be saved in /procesa/results/. The correspondence between results shown in paper and running scripts are shown in figure below.

index	config path	script path
1	procesa/configs/s2c2_esmc_cls/model1.py	procesa/scripts/s2c2_esmc_cls/s2c2_esmc-cls-model1.sh
2	procesa/configs/s2c2_esmc_cls/model2.py	procesa/scripts/s2c2_esmc_cls/s2c2_esmc-cls-model2.sh
3	procesa/configs/s2c5_esmc_cls/model1.py	procesa/scripts/s2c5_esmc_cls/s2c5_esmc-cls-model1.sh
4	procesa/configs/s2c5_esmc_cls/model3.py	procesa/scripts/s2c5_esmc_cls/s2c5_esmc-cls-model3.sh
5	procesa/configs/S_esmc_cls/model1.py	procesa/scripts/S_esmc_cls/S_esmc-cls-model1.sh
6	procesa/configs/S_esmc_cls/model0.py	procesa/scripts/S_esmc_cls/S_esmc-cls-model0.sh
7	procesa/configs/SC2_esmc_cls/model1.py	procesa/scripts/SC2_esmc_cls/SC2_esmc-cls-model1.sh
8	procesa/configs/SC2_esmc_cls/model2.py	procesa/scripts/SC2_esmc_cls/SC2_esmc-cls-model2.sh
9	procesa/configs/s2c2_esm1b/model112.py	procesa/scripts/s2c2_esm1b/slurm-esm1b-s2c2-model112.sh
10	procesa/configs/s2c2_esm1b/model90.py	procesa/scripts/s2c2_esm1b/slurm-esm1b-s2c2-model90.sh
11	procesa/configs/s2c2_esmc/model1.py	procesa/scripts/s2c2_esmc/s2c2_esmc-model1.sh
12	procesa/configs/s2c2_esmc/model3.py	procesa/scripts/s2c2_esmc/s2c2_esmc-model3.sh
13	procesa/configs/s2c5_esm1b/model104.py	procesa/scripts/s2c5_esm1b/slurm-esm1b-s2c5-model104.sh
14	procesa/configs/s2c5_esm1b/model89.py	procesa/scripts/s2c5_esm1b/slurm-esm1b-s2c5-model89.sh
15	procesa/configs/s2c5_esmc/model1.py	procesa/scripts/s2c5_esmc/s2c5_esmc-model1.sh
16	procesa/configs/s2c5_esmc/model0.py	procesa/scripts/s2c5_esmc/s2c5_esmc-model0.sh
17	procesa/configs/S_esm1b/model116.py	procesa/scripts/S_esm1b/S_esm1b-model116.sh
18	procesa/configs/S_esm1b/model115.py	procesa/scripts/S_esm1b/S_esm1b-model115.sh
19	procesa/configs/S_esmc/model1.py	procesa/scripts/S_esmc/S_esmc-model1.sh
20	procesa/configs/S_esmc/model3.py	procesa/scripts/S_esmc/S_esmc-model3.sh
21	procesa/configs/deepstabp_esmc/model0.py	procesa/scripts/deepstabp_esmc/deepstabp_esmc-model0.sh

Test

Example:

python test.py \
  configs/s2c2_esmc_cls/model1.py \
  --ckpt_path PATH-TO-YOUR-CKPT \
  --dataroot /datasets/procesa_data/hotprotein/ \
  --dataname s2c2_0 \
  --modelname esmc \
  --seed 101 \
  --task cls

Results

Results will be saved in /procesa/results folder.

Inference

Following /procesa/inference.ipynb to run procesa on single sequence.

Citations

If you make use of this code or the ProCeSa algorithm in your work, please cite the following paper:

@article{zhou2025procesa,
  title={ProCeSa: Contrast-Enhanced Structure-Aware Network for Thermostability Prediction with Protein Language Models},
  author={Zhou,  Feixiang and Zhang,  Shuo and Zhang,  Huifeng and Liu,  Jian K.},
  journal={Journal of Chemical Information and Modeling},
  year={2025},
  publisher={American Chemical Society (ACS)}
}