Accurate Prediction of Nanobody Structure:Molecular Modeling and Deep Learning

• 1、The Key Laboratory of Nonferrous Metal Materials and New Processing Technology of Ministry of Education, Guangxi University, Nanning 530004

Abstract：Antibodies are kind of important protein and a rapidly growing class of therapeutics. Recently, single domain camelid VHH antibodies, and their recognition nanobody domain(Nb) apperaed as a cost-effective highly stable based on accurate structural modeling of the variable domains that share a common fold and differ in the Comlementaryity Determining Regions (CDRs).The success of nanobodies is rooted in the simplicity and robustness of VHH scaffold and in its characteristic variability at the most variable CDR3 loop, which is frequently found to penetrate deeply into cavities of membrane protein targets. In this study, we tested the prediction accuracy of four methods for Nbs from the traditional physically based models methods and deep learning methods. In the way of physically based models, we used homologous modeling and Gaussian accelerated molecular dynamics (GaMD) simulation optimization methods; And Alphafold2 and RoseTTAFold, which have achieved remarkable results recently are the methods of deep learning. Comparing the prediction results of the four methods, we put forward prediction opinions for different types of Nbs, which is helpful to accurately prediction.

Keywords： CADD Structure prediction Nanobody Molecular dynamics simulation