What is hidden in the darkness? Characterization of AlphaFold structural space

Durairaj, Janani; Pereira, Joana; Akdel, Mehmet; Schwede, Torsten

doi:10.1101/2022.10.11.511548

Cited by 3 publications

(3 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, going from 1 to 100 query structures increases run time from 1.49 s to 3.13 s. When searching with multiple structures, most run time is in generating the query structure embeddings. Consequently, the speed benefits of the method arise when searching a structure or structures against the pre-computed embeddings of a huge database such as the AlphaFold database [7, 23]. By artificially copying the SCOPe search set to 10 million or 100 million structures we find that our method takes 2.7 s or 15.7 s to search this set with a single query on CPU.…”

Section: Resultsmentioning

confidence: 99%

“…Our method is similar to another approach that embeds protein structure [21], though our embedding is based on coordinates rather than hydrogen bonds. Embedding protein folds has also been done using residue-level features [22, 23], and GNNs acting on protein structure have been used for function prediction [24]. Other studies have used unsupervised contrastive learning on protein structures and show that the representations are useful for downstream prediction tasks including protein structural similarity [25, 26].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Fast protein structure searching using structure graph embeddings

Greener

Jamali

2022

Preprint

View full text Add to dashboard Cite

Comparing and searching protein structures independent of primary sequence has proved useful for remote homology detection, function annotation and protein classification. With the recent leap in accuracy of protein structure prediction methods and increased availability of protein models, attention is turning to how to best make use of this data. Fast and accurate methods to search databases of millions of structures will be essential to this endeavour, in the same way that fast protein sequence searching underpins much of bioinformatics. We train a simple graph neural network using supervised contrastive learning to learn a low-dimensional embedding of protein structure. The embedding can be used to search structures against large structural databases with accuracy comparable to current methods. The speed of the method and ability to scale to millions of structures makes it suitable for this structure-rich era. The method, called Progres, is available at https://github.com/jgreener64/progres.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Fast protein structure searching using structure graph embeddings

Greener

Jamali

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Because the models are intentionally topology-agnostic, we were also able to show that LRP can find important atoms from structures that exhibit the Urfold principle of “ architectural similarity despite topological variability ”—specifically, in this work the phosphate binding loops in Rossmann-fold proteins and P-Loop NTPases. In the future, we plan to identify and verify more common fragments and ancestral peptides by aligning and clustering ‘important’ regions from the cross-model fragments identified by LRP, while comparing them to known databases of potentially discontinuous fragment libraries (e.g., from shapemers [27], Fuzzle2.0 [28], ancestral peptides [14], themes [15], or TERMs [29]).…”

Section: Discussionmentioning

confidence: 99%

Explainable Deep Generative Models, Ancestral Fragments, and Murky Regions of the Protein Structure Universe

Draizen

Mura

Bourne

2022

Preprint

View full text Add to dashboard Cite

Modern proteins did not arise abruptly, as singular events, but rather over the course of at least 3.5 billion years of evolution. Can machine learning teach us how this occurred? The molecular evolutionary processes that yielded the intricate three-dimensional (3D) structures of proteins involve duplication, recombination and mutation of genetic elements, corresponding to short peptide fragments. Identifying and elucidating these ancestral fragments is crucial to deciphering the interrelationships amongst proteins, as well as how evolution acts upon protein sequences, structures & functions. Traditionally, structural fragments have been found using sequence-based and 3D structural alignment approaches, but that becomes challenging when proteins have undergone extensive permutations- allowing two proteins to share a common architecture, though their topologies may drastically differ (a phenomenon termed the Urfold). We have designed a new framework to identify compact, potentially-discontinuous peptide fragments by combining (i) deep generative models of protein superfamilies with (ii) layer-wise relevance propagation (LRP) to identify atoms of great relevance in creating an embedding during an all superfamilies by all domains analysis. Our approach recapitulates known relationships amongst the evolutionarily ancient small beta-barrels (e.g. SH3 and OB folds) and amongst P-loop-containing proteins (e.g. Rossmann and P-loop NTPases), previously established via manual analysis. Because of the generality of our deep model's approach, we anticipate that it can enable the discovery of new ancestral peptides. In a sense, our framework uses LRP as an 'explainable AI' approach, in conjunction with a recent deep generative model of protein structure (termed DeepUrfold), in order to leverage decades worth of structural biology knowledge to decipher the underlying molecular bases for protein structural relationships-including those which are exceedingly remote, yet discoverable via deep learning.

show abstract

Alternative Reading Frames are an Underappreciated Source of Protein Sequence Novelty

Ardern

2023

J Mol Evol

View full text Add to dashboard Cite

What is hidden in the darkness? Characterization of AlphaFold structural space

Cited by 3 publications

References 24 publications

Fast protein structure searching using structure graph embeddings

Fast protein structure searching using structure graph embeddings

Explainable Deep Generative Models, Ancestral Fragments, and Murky Regions of the Protein Structure Universe

Alternative Reading Frames are an Underappreciated Source of Protein Sequence Novelty

Contact Info

Product

Resources

About