netbiolab/LoGoPPI-Cross-species

LoGoPPI Cross-species

LoGoPPI is a protein–protein interaction prediction model built on ESM-2. It combines a Global sequence representation with a residue-level Maxsim score. The two calibrated scores are combined with fixed weights of 0.5 and 0.5.

Model overview

This checkpoint was trained on human protein interactions and evaluated on eight external species datasets. The Bernett checkpoint is trained separately and is available at LoGoPPI-Bernett.

Training and evaluation data

The human training data and the mouse, fly, worm, yeast, and E. coli test sets originate from the D-SCRIPT datasets. Three additional bacterial test sets are derived from STRING and are identified by taxonomy IDs 208964, 224308, and 83332.

The data/ directory contains the processed FASTA and query,text,label CSV files used for training, checkpoint selection, calibration, and evaluation. Test labels were used only for evaluation.

Results

Species	Pairs	AUPR
Fly	54,993	0.9166
Mouse	54,999	0.9131
Worm	54,996	0.8881
Yeast	54,963	0.7480
E. coli	18,232	0.7595
Taxonomy 208964	20,240	0.8568
Taxonomy 224308	18,722	0.8938
Taxonomy 83332	16,775	0.8526

Usage

Install LoGoPPI v2.0.0 and download the model files:

git clone https://github.com/netbiolab/LoGoPPI.git
cd LoGoPPI
conda env create -f environment.yml
conda activate logoppi

python - <<'PY'
from huggingface_hub import snapshot_download

snapshot_download(
    "netbiolab/LoGoPPI-Cross-species",
    revision="v2.0.0",
    local_dir="models/cross_species",
    ignore_patterns=["data/*"],
)
PY

Run inference with a FASTA file and a query,text pair CSV:

python inference.py \
  --model_dir models/cross_species \
  --fasta_path proteins.fasta \
  --pair_csv pairs.csv \
  --output_path predictions.csv \
  --gpus 0

Training, testing, sequence-only input, and embedding reuse are documented in the GitHub repository.

License and citation

LoGoPPI is released under the Apache License 2.0. Please cite the accompanying LoGoPPI study and the source datasets when using this model or the distributed data.