NbX: Machine Learning-Guided Re-Ranking of Nanobody-Antigen Binding Poses.

Chunlai Tam,Ashutosh Kumar,Kam Y. J. Zhang

doi:10.3390/ph14100968

Chunlai Tam, Ashutosh Kumar + Show 1 more

Open Access

https://doi.org/10.3390/ph14100968

Copy DOI

Abstract

Modeling the binding pose of an antibody is a prerequisite to structure-based affinity maturation and design. Without knowing a reliable binding pose, the subsequent structural simulation is largely futile. In this study, we have developed a method of machine learning-guided re-ranking of antigen binding poses of nanobodies, the single-domain antibody which has drawn much interest recently in antibody drug development. We performed a large-scale self-docking experiment of nanobody–antigen complexes. By training a decision tree classifier through mapping a feature set consisting of energy, contact and interface property descriptors to a measure of their docking quality of the refined poses, significant improvement in the median ranking of native-like nanobody poses by was achieved eightfold compared with ClusPro and an established deep 3D CNN classifier of native protein–protein interaction. We further interpreted our model by identifying features that showed relatively important contributions to the prediction performance. This study demonstrated a useful method in improving our current ability in pose prediction of nanobodies.

Highlights

Knowing the initial binding pose of an antibody (Ab) to its antigen (Ag) is required for in silico Ab affinity maturation and design
There are an emerging number of machine learning models that predict native complex of general protein–protein interaction [10,11,12], with some of them using popular neural network architectures in feature extraction from 3D coordinates of protein–protein complexes, such as 3D convolutional neural networks [13,14] and graph convolutional neural networks [15], which predict the nativeness of protein–protein interactions
We developed an Nb pose prediction model, NbX, and benchmarked its performance with ClusPro [2], which is one of the top performing protein–protein docking method from the latest CAPRI [30] and DOVE [14], a benchmarked binary classifier for native protein–protein interaction through deep 3D convolution

Summary

Introduction

Knowing the initial binding pose of an antibody (Ab) to its antigen (Ag) is required for in silico Ab affinity maturation and design. There are an emerging number of machine learning models that predict native complex of general protein–protein interaction [10,11,12], with some of them using popular neural network architectures in feature extraction from 3D coordinates of protein–protein complexes, such as 3D convolutional neural networks [13,14] and graph convolutional neural networks [15], which predict the nativeness of protein–protein interactions These models were able to improve the rankings of native protein–protein complexes from docking and existing re-ranking methods based on interface shape properties [16,17,18,19], evolutionary profile [20], physics and knowledge-based potentials [16,18,19,21,22,23]

Methods

Results

Conclusion