E(3)-invariant diffusion model for pocket-aware peptide generation

TL;DR

This work proposes a new method of computer-assisted inhibitor discovery: de novo pocket-aware peptide structure and sequence generation network that achieves comparable performance to state-of-the-art models, highlighting its potential in pocket-aware peptide design.

Abstract

Biologists frequently desire protein inhibitors for a variety of reasons, including use as research tools for understanding biological processes and application to societal problems in agriculture, healthcare, etc. Immunotherapy, for instance, relies on immune checkpoint inhibitors to block checkpoint proteins, preventing their binding with partner proteins and boosting immune cell function against abnormal cells. Inhibitor discovery has long been a tedious process, which in recent years has been accelerated by computational approaches. Advances in artificial intelligence now provide an opportunity to make inhibitor discovery smarter than ever before. While extensive research has been conducted on computer-aided inhibitor discovery, it has mainly focused on either sequence-to-structure mapping, reverse mapping, or bio-activity prediction, making it unrealistic for biologists to utilize such tools. Instead, our work proposes a new method of computer-assisted inhibitor discovery: de novo pocket-aware peptide structure and sequence generation network. Our approach consists of two sequential diffusion models for end-to-end structure generation and sequence prediction. By leveraging angle and dihedral relationships between backbone atoms, we ensure an E(3)-invariant representation of peptide structures. Our results demonstrate that our method achieves comparable performance to state-of-the-art models, highlighting its potential in pocket-aware peptide design. This work offers a new approach for precise drug discovery using receptor-specific peptide generation.

Figures (3)

• Figure 1: This figure illustrates the main architecture of our method, which includes: a) the architecture of the peptide structure prediction model, b) the architecture of the peptide sequence prediction model, and c) the architecture of a residue encoder (RE) layer. In Figures a and b, the "Self Attn" and "Cross Attn" blocks represent the self-attention and cross-attention mechanisms as proposed in the transformer research vaswani2017attention. In Figure c, "input" represents the input hidden state to be normalized, "Cond." denotes the condition, and "gate LN" refers to the gated layer normalization, which is defined by Equation \ref{['eq:gate_re']}
• Figure 2: This figure compares the distributions of generated angles with those from testing set. The Jensen-Shannon (JS) distance and Kullback-Leibler (KL) divergence are used to quantify the differences between two distributions. The top four plots illustrates the dihedral angle distributions, while the bottom plots displays the bond angle distributions.
• Figure 3: Ramachandran plot visualizes the distribution of secondary structure by comparing the $\phi$ and $\psi$ dihedral angles of amino acids. The areas corresponding to three main secondary structures -- namely $\beta$ sheet, right-handed (RH) $\alpha$ helix, and left-handed (LH) $\alpha$ helix -- are marked on the plot.