Unraveling the Role of Nanobodies Tetrad on Their Folding and Stability Assisted by Machine and Deep Learning Algorithms

• Matheus Vitor Ferreira Ferraz UFPE / Oswaldo Cruz Foundation http://orcid.org/0000-0002-6958-3115
• Wenny Camila dos Santos Adan Oswaldo Cruz Foundation http://orcid.org/0000-0002-4768-2200
• Roberto Dias Lins UFPE / Oswaldo Cruz Foundation http://orcid.org/0000-0002-3983-8025

Resumo

Nanobodies (Nbs) achieve high solubility and stability due to four conserved residues referred to as the Nb tetrad. While several studies have highlighted the importance of the Nbs tetrad to their stability, a detailed molecular picture of their role has not been provided. In this work, we have used the Rosetta package to engineer synthetic Nbs lacking the Nb tetrad and used the Rosetta Energy Function to assess the structural features of the native and designed Nbs concerning the presence of the Nb tetrad. To develop a classification model, we have benchmarked three different machine learning (ML) and deep learning (DL) algorithms and concluded that more complex models led to better binary classification for our dataset. Our results show that these two classes of Nbs differ significantly in features related to solvation energy and native-like structural properties. Notably, the loss of stability due to the tetrad’s absence is chiefly driven by the entropic contribution.

Palavras-chave: Camelid antibodies, Rosetta Energy Function, Machine learning