Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DualBind: A Dual-Loss Framework for Protein-Ligand Binding Affinity Prediction

Meng Liu, Saee Gopal Paliwal

TL;DR

DualBind is presented, a novel framework that integrates supervised mean squared error (MSE) with unsupervised denoising score matching (DSM) to accurately learn the binding energy function and improves generalizability and reduces reliance on labeled data compared to MSE-only models.

Abstract

Accurate prediction of protein-ligand binding affinities is crucial for drug development. Recent advances in machine learning show promising results on this task. However, these methods typically rely heavily on labeled data, which can be scarce or unreliable, or they rely on assumptions like Boltzmann-distributed data that may not hold true in practice. Here, we present DualBind, a novel framework that integrates supervised mean squared error (MSE) with unsupervised denoising score matching (DSM) to accurately learn the binding energy function. DualBind not only addresses the limitations of DSM-only models by providing more accurate absolute affinity predictions but also improves generalizability and reduces reliance on labeled data compared to MSE-only models. Our experimental results demonstrate that DualBind excels in predicting binding affinities and can effectively utilize both labeled and unlabeled data to enhance performance.

DualBind: A Dual-Loss Framework for Protein-Ligand Binding Affinity Prediction

TL;DR

DualBind is presented, a novel framework that integrates supervised mean squared error (MSE) with unsupervised denoising score matching (DSM) to accurately learn the binding energy function and improves generalizability and reduces reliance on labeled data compared to MSE-only models.

Abstract

Accurate prediction of protein-ligand binding affinities is crucial for drug development. Recent advances in machine learning show promising results on this task. However, these methods typically rely heavily on labeled data, which can be scarce or unreliable, or they rely on assumptions like Boltzmann-distributed data that may not hold true in practice. Here, we present DualBind, a novel framework that integrates supervised mean squared error (MSE) with unsupervised denoising score matching (DSM) to accurately learn the binding energy function. DualBind not only addresses the limitations of DSM-only models by providing more accurate absolute affinity predictions but also improves generalizability and reduces reliance on labeled data compared to MSE-only models. Our experimental results demonstrate that DualBind excels in predicting binding affinities and can effectively utilize both labeled and unlabeled data to enhance performance.
Paper Structure (6 sections, 7 equations, 3 figures, 2 tables)

This paper contains 6 sections, 7 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Boltzmann Distribution Assumption in DSM-only Models
  4. DualBind: A Dual-Loss Approach
  5. Experiments
  6. Conclusion

Figures (3)

  • Figure 1: (a) Distribution of binding affinity in the PDBbind v2020 refined dataset. It departs from the Boltzmann distribution. (b) Rank fit of a DSM-only model on training complexes.
  • Figure 2: An illustration of the DualBind methodology. DualBind employs a dual-loss framework that combines the MSE loss $\mathcal{L}_{\text{MSE}}$ and the DSM loss $\mathcal{L}_{\text{DSM}}$. Specifically, $\mathcal{L}_{\text{MSE}}$ anchors the predicted binding affinity of the crystal structure to its experimentally determined binding affinity. Concurrently, $\mathcal{L}_{\text{DSM}}$ shapes the gradient at the perturbed structure. Details are described in Section \ref{['sec:method']}.
  • Figure 3: Rank fit of DualBind on training complexes.