X-PAIR model checkpoints

This repository contains the trained model checkpoints used in X-PAIR, an ultrafast multitask framework for protein–protein interaction (PPI) and partner-specific interface prediction from protein sequences.

Paper Code Datasets Datasets

Model checkpoints

Checkpoint Task Training dataset
interaction_bernett.ckpt PPI prediction Gold Standard (Bernett et al. )
interaction_dscript.ckpt PPI prediction Cross-species benchmark (Sledzieski et al.)
interface_pioneer.ckpt Interface prediction PIONEER (Xiong et al.)
interaction_xfair.ckpt PPI prediction X-fair
interface_xfair.ckpt Interface prediction X-fair
multitask_xfair.ckpt PPI + interface prediction X-fair
multitask_xhuman.ckpt PPI + interface prediction X-human
multitask_xmultispecies.ckpt PPI + interface prediction X-multispecies

Model

X-PAIR requires Python ≥3.10 and can be installed from PyPI:

pip install xpair

The complete X-PAIR source code is publicly available on GitLab.

Installation and usage instructions are provided in the repository README.

Detailed documentation describing the software functionality, input data formats, model architecture, and available workflows is available in the X-PAIR documentation.

A minimal demo is provided for training, evaluation, and prediction.

Data

The exact processed datasets used to train and evaluate the models are publicly available on Hugging Face.

Datasets generated as part of the X-PAIR study are additionally archived on Zenodo.

Citation

If you use X-PAIR, please cite:

Rescalli, S. & Carbone, A.
X-PAIR: an ultrafast multitask framework for proteome-scale reconstruction of PPI networks and partner-specific interfaces from sequence.
bioRxiv (2026).
https://doi.org/10.64898/2026.07.20.739596

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support