Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evaluating protein binding interfaces with PUMBA

Azam Shirali, Giri Narasimhan

TL;DR

PuMba addresses the need for improved scoring in protein docking by replacing the PIsToN Vision Transformer backbone with Vision Mamba, enabling efficient long-range modeling of multi-channel interface features. The method preserves PIsToN's explainability framework while leveraging a bidirectional state-space architecture to better capture global and local patterns at protein interfaces. Across eight large public benchmarks, PuMba consistently outperforms PIsToN on metrics such as AUC ROC and average precision, and achieves higher docking-quality rankings and Top1 success rates. The work demonstrates that state-space modeled vision backbones are a viable and scalable alternative to transformers for biomolecular interface scoring, with open data and code for reproducibility.

Abstract

Protein-protein docking tools help in studying interactions between proteins, and are essential for drug, vaccine, and therapeutic development. However, the accuracy of a docking tool depends on a robust scoring function that can reliably differentiate between native and non-native complexes. PIsToN is a state-of-the-art deep learning-based scoring function that uses Vision Transformers in its architecture. Recently, the Mamba architecture has demonstrated exceptional performance in both natural language processing and computer vision, often outperforming Transformer-based models in their domains. In this study, we introduce PUMBA (Protein-protein interface evaluation with Vision Mamba), which improves PIsToN by replacing its Vision Transformer backbone with Vision Mamba. This change allows us to leverage Mamba's efficient long-range sequence modeling for sequences of image patches. As a result, the model's ability to capture both global and local patterns in protein-protein interface features is significantly improved. Evaluation on several widely-used, large-scale public datasets demonstrates that PUMBA consistently outperforms its original Transformer-based predecessor, PIsToN.

Evaluating protein binding interfaces with PUMBA

TL;DR

PuMba addresses the need for improved scoring in protein docking by replacing the PIsToN Vision Transformer backbone with Vision Mamba, enabling efficient long-range modeling of multi-channel interface features. The method preserves PIsToN's explainability framework while leveraging a bidirectional state-space architecture to better capture global and local patterns at protein interfaces. Across eight large public benchmarks, PuMba consistently outperforms PIsToN on metrics such as AUC ROC and average precision, and achieves higher docking-quality rankings and Top1 success rates. The work demonstrates that state-space modeled vision backbones are a viable and scalable alternative to transformers for biomolecular interface scoring, with open data and code for reproducibility.

Abstract

Protein-protein docking tools help in studying interactions between proteins, and are essential for drug, vaccine, and therapeutic development. However, the accuracy of a docking tool depends on a robust scoring function that can reliably differentiate between native and non-native complexes. PIsToN is a state-of-the-art deep learning-based scoring function that uses Vision Transformers in its architecture. Recently, the Mamba architecture has demonstrated exceptional performance in both natural language processing and computer vision, often outperforming Transformer-based models in their domains. In this study, we introduce PUMBA (Protein-protein interface evaluation with Vision Mamba), which improves PIsToN by replacing its Vision Transformer backbone with Vision Mamba. This change allows us to leverage Mamba's efficient long-range sequence modeling for sequences of image patches. As a result, the model's ability to capture both global and local patterns in protein-protein interface features is significantly improved. Evaluation on several widely-used, large-scale public datasets demonstrates that PUMBA consistently outperforms its original Transformer-based predecessor, PIsToN.
Paper Structure (9 sections, 2 figures, 3 tables)

This paper contains 9 sections, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Deep Learning in protein scoring functions
  4. Deep Learning in machine vision
  5. Explainability and hidden attention in ViM
  6. Our approach
  7. PuMba architecture
  8. Experimental Results
  9. Conclusions

Figures (2)

  • Figure 1: Overview of PIsToN and PuMba architectures: (\ref{['fig:arch_a']}, \ref{['fig:arch_b']}) Multi-attention architecture with hybrid feature processing in PIsToN and PuMba, (\ref{['fig:arch_c']}) bidirectional scans of patches of each feature in forward and backward SSM blocks in PuMba. The CLS token is inserted into the middle of the patch sequence to maximize bidirectional context utilization.
  • Figure 2: Explainability in PuMba with homodimer (PDB ID 1J3R, chains A and B).