A stacked meta-ensemble for protein inter-residue distance prediction

• Published in: Computers in biology and medicine 2022-09, Vol.148, p.105824-105824, Article 105824
• Main Authors: Rahman, Julia, Newton, M.A. Hakim, Hasan, Md. Al Mehedi, Sattar, Abdul
• Format: Article
• Language: English
• Subjects: Algorithms; Biology; Computers; Datasets; Deep learning; Ensemble learning; Inter-residue distance; Machine learning; Methods; Predictions; Protein structure; Protein structure prediction; Proteins; Real-valued distance; Residues; Search algorithms
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/j.compbiomed.2022.105824

Predicted inter-residue distances are a key behind recent success in high quality protein structure prediction (PSP). However, prediction of both short and long distance values together is challenging. Consequently, predicted short distances are mostly used by existing PSP methods. In this paper, we use a stacked meta-ensemble method to combine deep learning models trained for different ranges of real-valued distances. On five benchmark sets of proteins, our proposed inter-residue distance prediction method improves mean Local Distance Different Test (LDDT) scores at least by 5% over existing such methods. Moreover, using a real-valued distance based conformational search algorithm, we also show that predicted long distances help obtain significantly better protein conformations than when only predicted short distances are used. Our method is named meta-ensemble for distance prediction (MDP) and its program is available from https://gitlab.com/mahnewton/mdp. •A stacked meta-ensemble method to predict protein inter-residue long-range real-valued distances.•Improved Cβ−Cβ, Cα−Cα and N–O long-range real-valued distance prediction.•Improved protein 3D structure construction using long-range predicted distances.•The distance predictor’s program is available from https://gitlab.com/mahnewton/mdp.