Improving Antibody Design with Force-Guided Sampling in Diffusion Models

TL;DR

A novel approach to enhance the sampling process of diffusion models by integrating force field energy-based feedback, and demonstrates that this method guides the model to sample CDRs with lower energy, enhancing both the structure and sequence of the generated antibodies.

Abstract

Antibodies, crucial for immune defense, primarily rely on complementarity-determining regions (CDRs) to bind and neutralize antigens, such as viruses. The design of these CDRs determines the antibody's affinity and specificity towards its target. Generative models, particularly denoising diffusion probabilistic models (DDPMs), have shown potential to advance the structure-based design of CDR regions. However, only a limited dataset of bound antibody-antigen structures is available, and generalization to out-of-distribution interfaces remains a challenge. Physics based force-fields, which approximate atomic interactions, offer a coarse but universal source of information to better mold designs to target interfaces. Integrating this foundational information into diffusion models is, therefore, highly desirable. Here, we propose a novel approach to enhance the sampling process of diffusion models by integrating force field energy-based feedback. Our model, DiffForce, employs forces to guide the diffusion sampling process, effectively blending the two distributions. Through extensive experiments, we demonstrate that our method guides the model to sample CDRs with lower energy, enhancing both the structure and sequence of the generated antibodies.

Figures (7)

• Figure 1: The antigen-binding region comprises six complementarity-determining regions (CDRs). Each CDR is constructed from a variety of amino acids, which are themselves made up of atoms. These atoms are governed by forces, denoted by the symbol $F$.
• Figure 2: Antibody CDR generation with different sampling strategies. Upper: Standard DDPM sampling without force guidance. Lower: Incorporating force guidance into sampling, the model generates CDR structures with lower energy. Notation explained in the main text.
• Figure 3: Generated samples for the CDR-H3 region of the https://www.rcsb.org/structure/7DK2 antigen-antibody complex. The RMSD, binding energy ($\Delta\Delta G$), and amino acid sequences are reported. The antigen is in red, and the antibody in blue. All samples show improved binding over the reference structure.
• Figure 4: Energy of the https://www.rcsb.org/structure/7DK2 antigen-antibody complex's HCDR regions. Mean and standard error are based on $n = 25$ samples. The DiffForce converges to lower energy levels than DiffAb.
• Figure 5: Results for the https://www.rcsb.org/structure/7DK2 complex’s CDR-H3 region. Samples for DiffAb (top) and DiffForce (bottom) at timesteps $t = [15, 10, 5, 0]$. The energy and amino acid sequence are reported. DiffForce achieves better structure and lower energy earlier in the sampling.