Mechanism-Aware Protein-Protein Interaction Prediction via Contact-Guided Dual Attention on Protein Language Models:
git clone https://github.com/ChengfeiYan/PLMDA-PPI.git
Download the trained models from trained models.
For pair inference:
python predict.py sequenceA msaA pdbA sequenceB msaB pdbB result_path model_path device
1. sequenceA: fasta file corresponding to target A.
2. msaA: a3m file corresponding to target A (multiple sequence alignment).
3. pdbA: pdb file corresponding to target A.
4. sequenceB: fasta file corresponding to target B.
5. msaB: a3m file corresponding to target B (multiple sequence alignment).
6. pdbB: pdb file corresponding to target B.
7. result_path: [a directory for the output]
8. model_path: PLMDA-PPI(PDB) or PLMDA-PPI(Transfer)
9. device: cpu, cuda:0, cuda:1, ...
If you encounter that some residues in the pdb file are missing, you can use MODELLER to fill in these missing residues.
python predict.py 1Z6O_C.fasta 1Z6O_C.msa.a3m 1Z6O_C.pdb 1Z6O_O.fasta 1Z6O_O_msa.a3m 1Z6O_O.pdb result PLMDA-PPI(PDB).pt cpu
For batch-run inference:
python predict_list.py ppi_list.csv result_path model_path device
Where ppi_list.csv is a csv file of:
{protein_pair},{fasA},{a3mA},{pdbA},{fasB},{a3mB},{pdbB}
e.g.
1Z6O_C:1Z6O_O,1Z6O_C.fasta,1Z6O_C.msa.a3m,1Z6O_C.pdb,1Z6O_O.fasta,1Z6O_O_msa.a3m,1Z6O_O.pdb
The example test csv file is listed in the example directory.
The detailed script used to train PLMDA-PPI is in main_inter.py, which contains all the details of training PLMDA-PPI, including how to choose the best model, how to calculate the loss, etc.
For batch-run inference:
python train.py ppi_list.csv result_path device
Where ppi_list.csv is a csv file of:
{protein_pair},{len1},{len2},{fasA},{a3mA},{pdbA},{fasB},{a3mB},{pdbB},{interaction},{contact}
1. contact: txt file of true protein pair contact map.
e.g.
1Z6O_C:1Z6O_O,212,191,1Z6O_C.fasta,1Z6O_C.msa.a3m,1Z6O_C.pdb,1Z6O_O.fasta,1Z6O_O_msa.a3m,1Z6O_O.pdb,1,1Z6O_C_O.contact
The example train csv file is listed in the example directory.
Please cite: Mechanism-Aware Inductive Bias Enhances Generalization in Protein-Protein Interaction Prediction Shuchen Deng, Xuanjun Wan, Zichun Mu, Sheng-You Huang*, Chengfei Yan* bioRxiv 2025.07.04.663157; doi: https://doi.org/10.1101/2025.07.04.663157