Context-aware geometric deep learning for protein sequence design

• Published in: Nature communications 2024-07, Vol.15 (1), p.6273-10, Article 6273
• Main Authors: Krapp, Lucien F., Meireles, Fernando A., Abriata, Luciano A., Devillard, Jean, Vacle, Sarah, Marcaida, Maria J., Dal Peraro, Matteo
• Format: Article
• Language: English
• Subjects: 119/118; 631/114/469; 631/45/607; 631/61/338/469; 82/16; 82/6; 82/80; 82/83; Amino Acid Sequence; Amino acids; Concept learning; Datasets; Deep Learning; Design; Engineering; Enzymes; Humanities and Social Sciences; Ligands; Models, Molecular; multidisciplinary; Protein Conformation; Protein engineering; Protein Engineering - methods; Proteins; Proteins - chemistry; Proteins - metabolism; Science; Science (multidisciplinary); Sequences; Success; Transformers
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/s41467-024-50571-y

Protein design and engineering are evolving at an unprecedented pace leveraging the advances in deep learning. Current models nonetheless cannot natively consider non-protein entities within the design process. Here, we introduce a deep learning approach based solely on a geometric transformer of atomic coordinates and element names that predicts protein sequences from backbone scaffolds aware of the restraints imposed by diverse molecular environments. To validate the method, we show that it can produce highly thermostable, catalytically active enzymes with high success rates. This concept is anticipated to improve the versatility of protein design pipelines for crafting desired functions. Advances in deep learning are transforming protein design. Here, authors introduce a method using geometric transformers to predict protein sequences, resulting in highly thermostable and catalytically active enzymes with high success rates.