AlphaFold2 reveals commonalities and novelties in protein structure space for 21 model organisms

Abstract

Deep-learning (DL) methods like DeepMind’s AlphaFold2 (AF2) have led to substantial improvements in protein structure prediction. We analyse confident AF2 models from 21 model organisms using a new classification protocol (CATH-Assign) which exploits novel DL methods for structural comparison and classification. Of ~370,000 confident models, 92% can be assigned to 3253 superfamilies in our CATH domain superfamily classification. The remaining cluster into 2367 putative novel superfamilies. Detailed manual analysis on 618 of these, having at least one human relative, reveal extremely remote homologies and further unusual features. Only 25 novel superfamilies could be confirmed. Although most models map to existing superfamilies, AF2 domains expand CATH by 67% and increases the number of unique ‘global’ folds by 36% and will provide valuable insights on structure function relationships. CATH-Assign will harness the huge expansion in structural data provided by DeepMind to rationalise evolutionary changes driving functional divergence.

Assessing structure and disorder prediction tools for de novo emerged proteins in the age of machine learning

“…Also, entirely new structures are highly unlikely to derive from an ancestral protein homologous in sequence but structurally different [ Illergård et al, 2009 ; Chothia and Lesk, 1986 ]. Novel folds were also rarely identified within new experimentally solved structures [ Tóth-Petróczy and Tawfik, 2014 ] but recent advancements will increase dramatically the structural coverage of the known sequence space and could lead to identification and definition of new protein folds and families [ Liu et al, 2022 , Varadi et al, 2021 , Bordin et al, 2023 ]. These advancements also provide new opportunities to search for structural homology of de novo proteins on a larger set of protein structures with popular structure homology algorithms already including predictions [ van Kempen et al, 2022 ; La et al, 2009 ; Holm, 2022 ; Aderinwale et al, 2022 ].…”

“…All predictions have an average low (70> pLDDT >50) to very low (<50 pLDDT) confidence. Such a low confidence can be an indicator of disorder and/or of low-quality MSAs [Bordin et al, 2022]. Both disorder and lowquality MSAs are respectively a proposed property and a hallmark of de novo emerged proteins [Bornberg-Bauer et al, 2021].…”

What is hidden in the darkness? Deep-learning assisted large-scale protein family curation uncovers novel protein families and folds

“…Recently, several complementary approaches have been developed to categorise the diversity of the protein universe and uncover novelties (Akdel et al, 2022; Barrio-Hernandez et al, 2023; Bordin et al, 2023), again highlighting the importance of incorporating multiple perspectives and methods in protein function annotation. These approaches showcase the significance of using a diverse set of information to gain a more complete understanding of protein function and its role in cellular processes.…”

Does AlphaFold2 model proteins’ intracellular conformations? An experimental test using cross-linking mass spectrometry of endogenous ciliary proteins

“…While not the first attempt at proteome-wide structure prediction 9 , AF2's success stems from its high accuracy at ab initio prediction 1 . Its broad applicability across protein families without requiring prior structural knowledge has already led to the discovery of at least 26 entirely new protein folds 10 . Global benchmarking and independent validation of such predicted structures will be necessary to inform reliable and nuanced interpretations of these structures.…”