Protein Structure Determination using Metagenome sequence data

By Sergey Ovchinnikov, Hahnbeom Park, Neha J. Varghese, Po-Ssu Huang, Georgios A. Pavlopoulos, David E. Kim, Hetunandan Kamisetty, Nikos Kyrpides, David Baker. 2017

478 citations recorded by the source corpus.

Filling in the protein fold picture Fewer than a third of the 14,849 known protein families have at least one member with an experimentally determined structure. This leaves more than 5000 protein families with no structural information. Protein modeling using residue-residue contacts inferred from evolutionary data has been successful in modeling unknown structures, but it requires large numbers of aligned sequences. Ovchinnikov et al. augmented such sequence alignments with metagenome sequence data (see the Perspective by Söding). They determined the number of sequences required to allow modeling, developed criteria for model quality, and, where possible, improved modeling by matching predicted contacts to known structures. Their method predicted quality structural models for 614 protein families, of which about 140 represent newly discovered protein folds. Science, this issue p. 294; see also p. 248 Combining metagenome data with protein structure prediction generates models for 614 families with unknown structures. Despite decades of work by structural biologists, there are still ~5200 protein families with unknown structure outside the range of comparative modeling. We show that Rosetta structure prediction guided by residue-residue contacts inferred from evolutionary information can accurately model proteins that belong to large families and that metagenome sequence data more than triple the number of protein families with sufficient sequences for accurate modeling. We then integrate metagenome data, contact-based structure matching, and Rosetta structure calculations to generate models for 614 protein families with currently unknown structures; 206 are membrane proteins and 137 have folds not represented in the Protein Data Bank. This approach provides the representative models for large protein families originally envisioned as the goal of the Protein Structure Initiative at a fraction of the cost.

Subfields: Materials Chemistry, Molecular Biology, Computational Biology & Bioinformatics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Materials Chemistry

In Molecular Biology

In Computational Biology & Bioinformatics

Citations in the corpus

Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences (2019, background): “Contacts predicted from evolutionary data have enabled large-scale computational prediction of protein structures (Ovchinnikov et al., 2017).”
State-specific protein–ligand complex structure prediction with a multiscale deep generative model (2022, positive): “Specifically, evolutionary constraints extracted from multiple sequence alignments (MSAs) 2–5 or protein language models (PLMs) 6–9 and specialized neural networks are systematically combined with the sequence-based Article https://doi.org/10.1038/s42256-024-00792-z the physics-based methods…”
Improved protein structure prediction using predicted interresidue orientations (2019, positive): “We first evaluated the iterative hybridization protocol (23) previously used to improve models generated using direct contacts predicted from GREMLIN on the entire set of CASP13 and CAMEO targets (SI Appendix, Fig.”
Distance-based protein folding powered by deep learning (2018, methodology): “In CASP13, we did not make use of metagenomic sequence databases, which have been reported to be useful for some proteins (48).”
End-to-end differentiable learning of protein structure (2018, methodology): “A correct contact map can guide fragment assembly methods to an accurate 3D structure 25%– 50% of the time (Ovchinnikov et al., 2017).”
Clustering huge protein sequence sets in linear time (2017, background): “[22], who used an unpublished dataset of 2 billion metagenomic sequences.”
Enhancing Evolutionary Couplings with Deep Convolutional Neural Networks (2017, methodology): “Several tools (including Rosetta) have successfully incorporated evolutionary couplings into their pipelines as distance restraints to significantly improve predictions, particularly for proteins that have proven challenging using traditional approaches (Ovchinnikov et al., 2015, 2017).”
Folding membrane proteins by deep transfer learning (2017, methodology): “However, the CAMEO participants include some well-known servers such as Robetta(Kim et al., 2004) and its test version Server19(Ovchinnikov et al., 2017), Phyre(Kelley et al., 2015), RaptorX(Källberg et al., 2012), Swiss-Model(Biasini et al., 2014) and HHpred(Remmert et al., 2012).”
ProteinNet: a standardized data set for machine learning of protein structure (2019, background): “metagenomic sequences from the Joint Genome Institute [22] and filtering to include only sequences available prior to CASP start dates (Table 2).”
Prediction of interresidue contacts with DeepMetaPSICOV in CASP13 (2019, methodology): “It has been established that metagenomic sequence collections are a rich source of sequence data that can be used for this purpose.(24) Therefore, in CASP13, we improved upon our previous approach for generating deeper MSAs(5,25) by including both UniRef100 and metagenomic sequences in the search.”
Assessment of hard target modeling in CASP12 reveals an emerging role of alignment‐based contact prediction methods (2018, methodology): “In fact, processing the T0866 sequence with the EVFold, 13 Gremlin, 14 and RaptorX 15 web-servers returns very clear coevolution patterns that match quite well the intra-and inter-monomer contact maps (Figure S1A).”
Blind prediction of homo- and hetero-protein complexes: The CASP13-CAPRI experiment (2019, neutral): “10-12 Owing to the recent explosion of the number of available protein sequences, and progress in computational methods for exploiting them to predict residue-residue contacts, 13-15 the ability to model protein structures in absence of available templates has also made significant strides forward.”