Predict metal-binding proteins and structures through integration of evolutionary-scale and physics-based modeling DOI Creative Commons
Xin Dai,

Max Henderson,

Shinjae Yoo

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Авг. 10, 2024

ABSTRACT Metals are essential elements in all living organisms, binding to approximately 50% of proteins. They serve stabilize proteins, catalyze reactions, regulate activities, and fulfill various physiological pathological functions. While there have been many advancements determining the structures protein-metal complexes, numerous metal-binding proteins still need be identified through computational methods validated experiments. To address this need, we developed ESMBind workflow, which combines evolutionary scale modeling (ESM) for prediction physics-based modeling. Our approach utilizes ESM-2 ESM-IF models predict probability at residue level. In addition, designed a metal-placement method energy minimization technique generate detailed 3D complexes. workflow outperforms other terms 3D-level predictions. demonstrate its effectiveness, applied 142 uncharacterized fungal pathogen predicted involved infection virulence.

Язык: Английский

Predicting metal-binding proteins and structures through integration of evolutionary-scale and physics-based modeling DOI
Xin Dai,

Max Henderson,

Shinjae Yoo

и другие.

Journal of Molecular Biology, Год журнала: 2025, Номер unknown, С. 168962 - 168962

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

2
MERIT: Accurate prediction of multi ligand-binding residues with hybrid deep transformer network, evolutionary couplings and transfer learning DOI
Jian Zhang, Sushmita Basu, Fuhao Zhang

и другие.

Journal of Molecular Biology, Год журнала: 2024, Номер unknown, С. 168872 - 168872

Опубликована: Ноя. 1, 2024

Язык: Английский

Процитировано

3
bindNode24: Competitive binding residue prediction with 60% smaller model DOI Creative Commons
Kyra Erckert,

Franz Birkeneder,

Burkhard Rost

и другие.

Computational and Structural Biotechnology Journal, Год журнала: 2025, Номер 27, С. 1060 - 1066

Опубликована: Янв. 1, 2025

Many proteins function through ligand binding. Yet, reliable experimental binding data remains limited. Recent advances predict residues from sequences using protein Language Model embeddings. The AlphaFold Protein Structure Database, which has 3D structure predictions AlphaFold2, opens the way for graph neural networks that residues. Here, we introduce bindNode24, a new method Graph Neural Networks to whether residue binds any of three classes: small molecules, metal ions, and nucleic macromolecules. Compared state-of-the-art, this approach reduces number free parameters by almost 60 % at similar performance. Our findings also suggest secondary tertiary features AlphaFold2 are easy integrate into prediction tasks previously solely relied on

Язык: Английский

Процитировано

0
Predicting metal-protein interactions using cofolding methods: Status quo DOI Creative Commons
Simon Dürr, Ursula Röthlisberger

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Июнь 2, 2024

Abstract Metals play important roles for enzyme function and many therapeutically relevant proteins. Despite the fact that first drugs developed via computer aided drug design were metalloprotein inhibitors, computational pipelines discovery still discard metalloproteins due to difficulties of modelling them computationally. New “cofolding” methods such as AlphaFold3 (AF3) ( Abramson et al., 2024 ) RoseTTAfold-AllAtom (RFAA) Krishna promise improve this issue by being able dock small molecules in presence multiple complex cofactors including metals or covalent modifications. Here, we analyze current status metal ion prediction using these methods. We find currently only AF3 provides realistic predictions ions, RFAA contrast does perform worse than more specialized models AllMetal3D predicting location ions accurately. are consistent with expected physico-chemical trends/intuition whereas often also predicts unrealistic locations.

Язык: Английский

Процитировано

2
MetaLATTE: Metal Binding Prediction via Multi-Task Learning on Protein Language Model Latents DOI Creative Commons
Yinuo Zhang,

Phil He,

Ashley Hsu

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Июнь 29, 2024

Abstract The bioremediation of environments contaminated with heavy metals is an important challenge in environmental biotechnology, which may benefit from the identification proteins that bind and neutralize these metals. Here, we introduce a novel predictive algorithm conducts Metal binding prediction via LA nguage model la T en E mbeddings using multi-task learning approach to accurately classify metal-binding properties input protein sequences. Our MetaLATTE utilizes state-of-the-art ESM-2 language (pLM) embeddings position-sensitive attention mechanism predict likelihood specific metals, such as zinc, lead, mercury. Importantly, our addresses challenges posed by understudied organisms, are often absent traditional databases, without requirement structure. By providing probability distribution over potential classifier elucidates interactions diverse metal ions. We envision will serve powerful tool for rapidly screening identifying new proteins, metagenomic discovery or de novo design efforts, can later be employed targeted campaigns.

Язык: Английский

Процитировано

0
Predict metal-binding proteins and structures through integration of evolutionary-scale and physics-based modeling DOI Creative Commons
Xin Dai,

Max Henderson,

Shinjae Yoo

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Авг. 10, 2024

ABSTRACT Metals are essential elements in all living organisms, binding to approximately 50% of proteins. They serve stabilize proteins, catalyze reactions, regulate activities, and fulfill various physiological pathological functions. While there have been many advancements determining the structures protein-metal complexes, numerous metal-binding proteins still need be identified through computational methods validated experiments. To address this need, we developed ESMBind workflow, which combines evolutionary scale modeling (ESM) for prediction physics-based modeling. Our approach utilizes ESM-2 ESM-IF models predict probability at residue level. In addition, designed a metal-placement method energy minimization technique generate detailed 3D complexes. workflow outperforms other terms 3D-level predictions. demonstrate its effectiveness, applied 142 uncharacterized fungal pathogen predicted involved infection virulence.

Язык: Английский

Процитировано

0