Publications

Protein–protein interaction (PPI) predictors are often evaluated with random pair splits that permit proteins to recur across training and test sets, while constructed negative pairs can introduce a separate source of benchmark-specific signal. We reassessed the Guo yeast benchmark using a matched protein-novelty curve, independently generated MMseqs2 sequence-cluster-controlled R3 variants, topology-only controls, degree- and length-matched putative negatives, handcrafted and frozen protein-language-model representations, multiple model configurations, seed analyses, and connected-component-aware inference. In the matched R1→R2→R3 comparison, AAC+CTD–MLP ROC-AUC declined from 0.697 for seen–seen pairs to 0.626 for one-unseen pairs and 0.587 for both-unseen pairs. Across independently regenerated cluster-disjoint R3 variants at 50%, 40%, 30%, and 20% MMseqs2 clustering thresholds, performance showed no consistent monotonic trend. Guo original-negative endpoints had lower positive-network degree than positive endpoints (mean 3.34 versus 6.72), and a degree-only logistic model reached ROC-AUC 0.962 on the original negatives but 0.502 on degree-matched negatives. We used degree-matched negatives consistently for R3 training and testing; six sequence-based model configurations yielded ROC-AUC 0.493–0.547; none showed corrected connected-component-aware evidence of above-chance discrimination. A PIPR multispecies analysis reproduced degree-associated label separation but also exposed cross-species pairing as an additional pooled cue. These results show that protein novelty, operational sequence-cluster separation, and negative-set topology are distinct dimensions of PPI benchmark evaluation. The Guo result is benchmark-specific and should not be generalized to sequence-based PPI prediction or protein language models overall. The dataset and source code for the experiments are publicly available at https://github.com/morufwork/guoAudit.git . Random pair splits, protein-disjoint splits, sequence-cluster controls, and negative-set controls answer different evaluation questions and should be reported separately. In Guo, AAC+CTD–MLP ROC-AUC declined from 0.697 at R1 to 0.587 at R3, while independently generated R3-50/40/30/20 variants showed no monotonic threshold trend. Guo original negatives carried strong positive-network-degree imbalance; a degree-only logistic model achieved ROC-AUC 0.962 on N0 but 0.502 on degree-matched N2. Under consistent R3/N2 evaluation, six sequence-model configurations had ROC-AUC 0.493–0.547, and the secondary PIPR analysis showed that pooled degree matching can introduce an additional taxonomic-pairing cue.

Protein–protein interactions (PPIs) play a central role in elucidating cellular mechanisms. However, a substantial gap remains in current prediction models, as they frequently overlook the structural and physicochemical context governing molecular binding, thereby limiting predictive accuracy. To address this limitation, we introduce predPPI-GReMLIN, a graph-based framework that represents protein–protein interfaces as bipartite graphs, integrating atomic-level physicochemical descriptors, spatial distance constraints, and conserved substructure mining to perform PPI prediction. The method employs a graph-search strategy to detect interface interaction patterns and conducts ligand swapping with complexes that share the same patterns as the query, enabling the prediction of novel interaction partners. Evaluations across multiple datasets—including CAMP (protein–peptide), Yeast (binary PPI), and TAGPPI (multi-class PPI)—demonstrate consistently strong predictive performance, achieving precision, recall, accuracy, and F1-scores exceeding 97% on binary classification benchmarks and surpassing state-of-the-art sequence- and structure-based approaches. Furthermore, incorporating solvent-accessible surface area (SASA)-derived features improved multi-class interaction-type classification accuracy to 57.74%. A case study on the SARS-CoV-2 spike–ACE2 complex further validated the approach: docking simulations using ligands predicted by our method reproduced native-like binding energetics comparable to redocking results. Additionally, a comparison between docking using a predPPI-GReMLIN-predicted ligand and docking using ligands selected at random from the PDB yielded a highly significant p-value (p = 5.1 × 10⁻⁹), indicating a robust statistical difference in binding performance. Collectively, these findings demonstrate predPPI-GReMLIN’s ability to capture conserved structural determinants of PPIs, providing a robust and interpretable framework for protein interaction prediction and ligand discovery. The dataset and source code for the experiments are publicly available at https://github.com/morufwork/predPPIGReMLIN.git.