SWORD2: hierarchical analysis of protein 3D structures

Cretin, Gabriel; Galochkina, Tatiana; Meersche, Yann Vander; Brevern, Alexandre G. de; Postic, Guillaume; Gelly, Jean‐Christophe

doi:10.1093/nar/gkac370

Cited by 18 publications

(11 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model presented here is trained and validated on protein domains; due to the domain-specific nature of the training it is not expected to work without modification on protein chains containing multiple domains, long disordered regions or complexes. Fortunately, there are a number of tools such as Merizo [50], SWORD2 [51] and Chainsaw [52] that can split query structures into domains. Searching with domains separately can overcome issues that arise from searching with multiple domains at the same time, such as missing related proteins due to differing orientations of the domains.…”

Section: Discussionmentioning

confidence: 99%

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: Discussionmentioning

confidence: 99%

Fast protein structure searching using structure graph embeddings

Greener

Jamali

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…While Birch has performed well through the development stage of Slice’N’Dice , large scale testing will now be undertaken to identify which clustering algorithm performs best. Future work might additionally look into the inclusion of other clustering methods such as SWORD2 (Cretin et al ., 2022) and DCI (Kumar et al ., 2022), and the combination of clustering methods in a consensus strategy. We’re also aware that clustering in combination with the removal of low confidence residues can occasionally leave disconnected fragments (Fig.…”

Section: Discussionmentioning

confidence: 99%

Slice’N’Dice: Maximising the value of predicted models for structural biologists

Simpkin

Elliott

Stevenson

et al. 2022

Preprint

View full text Add to dashboard Cite

With the advent of next generation modelling methods, such as AlphaFold2, structural biologists are increasingly using predicted structures as search models for Molecular Replacement (MR) when experimental structures of homologues are unavailable. Inaccuracy in domain-domain orientations is often a key limitation when using predicted models for MR. Slice'N'Dice is a software package designed to address this issue by first slicing models into distinct structural units and then automatically placing the slices using Phaser. The slicing step can use AlphaFold2's predicted aligned error (PAE), or can operate via a variety of Cα atom clustering algorithms, extending applicability to structures of any origin. The number of splits can be selected by the user. Slice'N'Dice is available in CCP4 8.0 and is currently being adapted for cryo-EM use cases.

show abstract

“…The 3D structure predictions of the proteins that demonstrated a significant hit with the PD fold database were used to identify corresponding (so-called) protein units: stable segments within a protein that fall between secondary structures and domains in terms of their complexity. These were determined with SWORD2, which analyzes the spatial relationships between α-carbon atoms and optimizes for structural autonomy within these segments (Cretin et al 2022). Every protein unit identified was subsequently screened against the PD fold database to locate the best match.…”

Section: Methodsmentioning

confidence: 99%

DepoScope: accurate phage depolymerase annotation and domain delineation using large language models

Concha-Eloko,

Stock,

De Baets

et al. 2024

Preprint

View full text Add to dashboard Cite

Bacteriophages (phages) are viruses that infect bacteria. Many of them produce specific enzymes called depolymerases to break down external polysaccharide structures. Accurate annotation and domain identification of these depolymerases are challenging due to their inherent sequence diversity. Hence, we present DepoScope, a machine learning tool that combines a fine-tuned ESM-2 model with a convolutional neural network to precisely identify depolymerase sequences and their enzymatic domains. To accomplish this, we curated a dataset from the INPHARED phage genome database, created a polysaccharide-degrading domain database, and applied sequential filters to construct a high-quality dataset, which are subsequently used to train DepoScope. Our work is the first approach that combines sequence-level predictions with amino-acid-level predictions for an accurate depolymerase detection and functional domain identification. In that way, we believe that DepoScope can enhance our understanding of phage-host interactions at the level of depolymerases.Summary with Key MessagesPhage depolymerases are proteins that play a crucial role in the first step of a phage replication cycle. As a result, they are both important from a biological perspective and a therapeutical perspective.Current methods to accurately annotate phage depolymerases and their associated enzymatic domains remains challenging due to their inherent high sequence diversity.We have developed DepoScope, a language-based artificial intelligence model that can accurately identify phage depolymerases and their specific enzymatic domains.We provide full public access to the DepoScope code and database to give broad access to the research community and promote further research.

show abstract

SWORD2: hierarchical analysis of protein 3D structures

Cited by 18 publications

References 39 publications

Fast protein structure searching using structure graph embeddings

Fast protein structure searching using structure graph embeddings

Slice’N’Dice: Maximising the value of predicted models for structural biologists

DepoScope: accurate phage depolymerase annotation and domain delineation using large language models

Contact Info

Product

Resources

About