What does this research mean for the field?

The PTP-SMGCA framework significantly improves peptide toxicity prediction accuracy, achieving an AUROC of 0.9289 and an AUPRC of 0.9397. Novelty: ClaimNovelty.NOVEL_FINDING. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To develop a high-precision and scalable framework for peptide toxicity prediction.

March 5, 2026Journal of Chemical Information and Modeling

Dual-Modal Fusion Based on Peptide Sequence and Molecular Graph with Bidirectional Cross-Attention for Precise Peptide Toxicity Prediction

View Full Paper

Ask AI

Bookmark

Authors

PLPan LiUniversity of Science and Technology Liaoning SZShaopeng ZhangUniversity of Science and Technology Liaoning RLRan LiuUniversity of Science and Technology Liaoning

Discussion

Member takes

Implication

This framework improves peptide toxicity prediction, suggesting a new approach for therapeutic peptide drug development.

Key Points

To develop a high-precision and scalable framework for peptide toxicity prediction.
Proposed a dual-modal feature fusion framework called PTP-SMGCA.
Utilized bidirectional cross-attention for aligning molecular graph and peptide sequence features.
Captured both local motifs and long-range dependencies in peptide sequences and molecular graphs.
Achieved an AUROC of 0.9289 and an AUPRC of 0.9397 for toxicity prediction.
Outperformed several leading toxicity prediction tools such as ATSE, CAPTP, and ToxiPep.
Identified key amino acids that contribute to toxicity through interpretability analysis.

Cite This Study

Li et al. (2026) studied this question.

synapsesocial.com/papers/69a91cbed6127c7a504bfa58 https://doi.org/10.1021/acs.jcim.6c00033

View Full Paper

Ask AI

Bookmark