Direct prediction of intrinsically disordered protein conformational properties from sequence

Nature Methods, Год журнала: 2024, Номер 21(3), С. 465 - 476

Опубликована: Янв. 31, 2024

Abstract Intrinsically disordered regions (IDRs) are ubiquitous across all domains of life and play a range functional roles. While folded generally well described by stable three-dimensional structure, IDRs exist in collection interconverting states known as an ensemble. This structural heterogeneity means that largely absent from the Protein Data Bank, contributing to lack computational approaches predict ensemble conformational properties sequence. Here we combine rational sequence design, large-scale molecular simulations deep learning develop ALBATROSS, deep-learning model for predicting dimensions IDRs, including radius gyration, end-to-end distance, polymer-scaling exponent asphericity, directly sequences at proteome-wide scale. ALBATROSS is lightweight, easy use accessible both locally installable software package point-and-click-style interface via Google Colab notebooks. We first demonstrate applicability our predictors examining generalizability sequence–ensemble relationships IDRs. Then, leverage high-throughput nature characterize sequence-specific biophysical behavior within between proteomes.

Язык: Английский

---

High-throughput prediction of protein conformational distributions with subsampled AlphaFold2

Nature Communications, Год журнала: 2024, Номер 15(1)

Опубликована: Март 27, 2024

Abstract This paper presents an innovative approach for predicting the relative populations of protein conformations using AlphaFold 2, AI-powered method that has revolutionized biology by enabling accurate prediction structures. While 2 shown exceptional accuracy and speed, it is designed to predict proteins’ ground state limited in its ability conformational landscapes. Here, we demonstrate how can directly different subsampling multiple sequence alignments. We tested our against nuclear magnetic resonance experiments on two proteins with drastically amounts available data, Abl1 kinase granulocyte-macrophage colony-stimulating factor, predicted changes their more than 80% accuracy. Our worked best when used qualitatively effects mutations or evolution landscape well-populated states proteins. It thus offers a fast cost-effective way at even single-point mutation resolution, making useful tool pharmacology, analysis experimental results, evolution.

Язык: Английский

---

Approximating Projections of Conformational Boltzmann Distributions with AlphaFold2 Predictions: Opportunities and Limitations

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(3), С. 1434 - 1447

Опубликована: Янв. 12, 2024

Protein thermodynamics is intimately tied to biological function and can enable processes such as signal transduction, enzyme catalysis, molecular recognition. The relative free energies of conformations that contribute these functional equilibria evolved for the physiology organism. Despite importance understanding developing treatments disease, computational experimental methods capable quantifying energetic determinants are limited systems modest size. Recently, it has been demonstrated artificial intelligence system AlphaFold2 be manipulated produce structurally valid protein conformational ensembles. Here, we extend studies explore extent which contact distance distributions approximate projections Boltzmann distributions. For this purpose, examine joint probability inter-residue distances along functionally relevant collective variables several systems. Our suggest normalized correlate with conformation probabilities obtained other but they suffer from peak broadening. We also find sensitive point mutations. Overall, anticipate our findings will valuable community seeks model changes in large biomolecular

Язык: Английский

---

AlphaFold2 and Deep Learning for Elucidating Enzyme Conformational Flexibility and Its Application for Design

JACS Au, Год журнала: 2023, Номер 3(6), С. 1554 - 1562

Опубликована: Июнь 6, 2023

The recent success of AlphaFold2 (AF2) and other deep learning (DL) tools in accurately predicting the folded three-dimensional (3D) structure proteins enzymes has revolutionized structural biology protein design fields. 3D indeed reveals key information on arrangement catalytic machinery which elements gate active site pocket. However, comprehending enzymatic activity requires a detailed knowledge chemical steps involved along cycle exploration multiple thermally accessible conformations that adopt when solution. In this Perspective, some studies showing potential AF2 elucidating conformational landscape are provided. Selected examples developments AF2-based DL methods for discussed, as well few enzyme cases. These show allowing routine computational efficient enzymes.

Язык: Английский

---

Drug specificity and affinity are encoded in the probability of cryptic pocket opening in myosin motor domains

eLife, Год журнала: 2023, Номер 12

Опубликована: Янв. 27, 2023

The design of compounds that can discriminate between closely related target proteins remains a central challenge in drug discovery. Specific therapeutics targeting the highly conserved myosin motor family are urgently needed as mutations at least six its members cause numerous diseases. Allosteric modulators, like myosin-II inhibitor blebbistatin, promising means to achieve specificity. However, it unclear why blebbistatin inhibits motors with different potencies given binds pocket is always closed blebbistatin-free experimental structures. We hypothesized probability opening an important determinant potency blebbistatin. To test this hypothesis, we used Markov state models (MSMs) built from over 2 ms aggregate molecular dynamics simulations explicit solvent. find blebbistatin's binding readily opens blebbistatin-sensitive isoforms. Comparing these conformational ensembles reveals correctly identifies which isoforms most sensitive inhibition and docking against MSMs quantitatively predicts affinities (R

Язык: Английский

---

Artificial intelligence in small molecule drug discovery from 2018 to 2023: Does it really work?

Bioorganic Chemistry, Год журнала: 2023, Номер 141, С. 106894 - 106894

Опубликована: Сен. 27, 2023

Язык: Английский

---

Exploring Kinase Asp-Phe-Gly (DFG) Loop Conformational Stability with AlphaFold2-RAVE

Journal of Chemical Information and Modeling, Год журнала: 2023, Номер 64(7), С. 2789 - 2797

Опубликована: Ноя. 20, 2023

Kinases compose one of the largest fractions human proteome, and their misfunction is implicated in many diseases, particular, cancers. The ubiquitousness structural similarities kinases make specific effective drug design difficult. In conformational variability due to evolutionarily conserved Asp-Phe-Gly (DFG) motif adopting out conformations relative stabilities thereof are key structure-based for ATP competitive drugs. These extremely sensitive small changes sequence provide an important problem sampling method development. Since invention AlphaFold2, world has noticeably changed. spite it being limited crystal-like structure prediction, several methods have also leveraged its underlying architecture improve dynamics enhanced ensembles, including AlphaFold2-RAVE. Here, we extend AlphaFold2-RAVE apply a set kinases: wild type DDR1 three mutants with single point mutations that known behave drastically differently. We show able efficiently recover stability using transferable learned order parameters potentials, thereby supplementing AlphaFold2 as tool exploration Boltzmann-weighted protein (Meller, A.; Bhakat, S.; Solieva, Bowman, G. R. Accelerating Cryptic Pocket Discovery Using AlphaFold. J. Chem. Theory Comput. 2023, 19, 4355–4363).

Язык: Английский

---

Protein Ensemble Generation Through Variational Autoencoder Latent Space Sampling

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(7), С. 2689 - 2695

Опубликована: Март 28, 2024

Mapping the ensemble of protein conformations that contribute to function and can be targeted by small molecule drugs remains an outstanding challenge. Here, we explore use variational autoencoders for reducing challenge dimensionality in structure generation problem. We convert high-dimensional structural data into a continuous, low-dimensional representation, carry out search this space guided quality metric, then RoseTTAFold sampled information generate 3D structures. approach ensembles cancer relevant K-Ras, train VAE on subset available K-Ras crystal structures MD simulation snapshots, assess extent sampling close withheld from training. find our latent procedure rapidly generates with high is able sample within 1 Å held-out structures, consistency higher than or AlphaFold2 prediction. The sufficiently recapitulate cryptic pockets allow docking.

Язык: Английский

---

Deep learning for protein structure prediction and design—progress and applications

Molecular Systems Biology, Год журнала: 2024, Номер 20(3), С. 162 - 169

Опубликована: Янв. 30, 2024

Abstract Proteins are the key molecular machines that orchestrate all biological processes of cell. Most proteins fold into three-dimensional shapes critical for their function. Studying 3D shape can inform us mechanisms underlie in living cells and have practical applications study disease mutations or discovery novel drug treatments. Here, we review progress made sequence-based prediction protein structures with a focus on go beyond single monomer structures. This includes application deep learning methods complexes, different conformations, evolution these to design. These developments create new opportunities research will impact across many areas biomedical research.

Язык: Английский

---

Toward physics‐based precision medicine: Exploiting protein dynamics to design new therapeutics and interpret variants

Protein Science, Год журнала: 2024, Номер 33(3)

Опубликована: Фев. 15, 2024

The goal of precision medicine is to utilize our knowledge the molecular causes disease better diagnose and treat patients. However, there a substantial mismatch between small number food drug administration (FDA)-approved drugs annotated coding variants compared needs medicine. This review introduces concept physics-based medicine, scalable framework that promises improve understanding sequence-function relationships accelerate discovery. We show accounting for ensemble structures protein adopts in solution with computer simulations overcomes many limitations imposed by assuming single structure. highlight studies dynamics recent methods analysis structural ensembles. These demonstrate differences conformational distributions predict functional within families variants. Thanks new computational tools are providing unprecedented access ensembles, this insight may enable accurate predictions variant pathogenicity entire libraries further explicitly like alchemical free energy calculations or docking Markov state models, can uncover novel lead compounds. To conclude, we cryptic pockets, cavities absent experimental structures, provide an avenue target proteins currently considered undruggable. Taken together, provides roadmap field science

Язык: Английский

---