Scientific discovery in the age of artificial intelligence

Nature, Journal Year: 2023, Volume and Issue: 620(7972), P. 47 - 60

Published: Aug. 2, 2023

Language: Английский

---

Recent advances and applications of deep learning methods in materials science

npj Computational Materials, Journal Year: 2022, Volume and Issue: 8(1)

Published: April 5, 2022

Deep learning (DL) is one of the fastest growing topics in materials data science, with rapidly emerging applications spanning atomistic, image-based, spectral, and textual modalities. DL allows analysis unstructured automated identification features. Recent development large databases has fueled application methods atomistic prediction particular. In contrast, advances image spectral have largely leveraged synthetic enabled by high quality forward models as well generative unsupervised methods. this article, we present a high-level overview deep-learning followed detailed discussion recent developments deep simulation, imaging, analysis, natural language processing. For each modality discuss involving both theoretical experimental data, typical modeling approaches their strengths limitations, relevant publicly available software datasets. We conclude review cross-cutting work related to uncertainty quantification field brief perspective on challenges, potential growth areas for science. The science presents an exciting avenue future discovery design.

Language: Английский

---

Machine learning in protein structure prediction

Current Opinion in Chemical Biology, Journal Year: 2021, Volume and Issue: 65, P. 1 - 8

Published: May 18, 2021

Prediction of protein structure from sequence has been intensely studied for many decades, owing to the problem's importance and its uniquely well-defined physical computational bases. While progress historically ebbed flowed, past two years saw dramatic advances driven by increasing "neuralization" prediction pipelines, whereby computations previously based on energy models sampling procedures are replaced neural networks. The extraction contacts evolutionary record; distillation sequence-structure patterns known structures; incorporation templates homologs in Protein Databank; refinement coarsely predicted structures into finely resolved ones have all reformulated using Cumulatively, this transformation resulted algorithms that can now predict single domains with a median accuracy 2.1 Å, setting stage foundational reconfiguration role biomolecular modeling within life sciences.

Language: Английский

---

The transformational role of GPU computing and deep learning in drug discovery

Nature Machine Intelligence, Journal Year: 2022, Volume and Issue: 4(3), P. 211 - 221

Published: March 23, 2022

Language: Английский

---

Bottom-up Coarse-Graining: Principles and Perspectives

Journal of Chemical Theory and Computation, Journal Year: 2022, Volume and Issue: 18(10), P. 5759 - 5791

Published: Sept. 7, 2022

Large-scale computational molecular models provide scientists a means to investigate the effect of microscopic details on emergent mesoscopic behavior. Elucidating relationship between variations scale and macroscopic observable properties facilitates an understanding interactions driving real world materials complex systems (e.g., those found in biology, chemistry, science). As result, discovering explicit, systematic connection nature behavior is fundamental goal for this type investigation. The forces critical heterogeneous are often unclear. More problematically, simulations representative model prohibitively expensive from both spatial temporal perspectives, impeding straightforward investigations over possible hypotheses characterizing While reduction resolution study, such as moving atomistic simulation that large coarse-grained (CG) groups atoms, can partially ameliorate cost individual simulations, proposed intermediate nontrivial presents new obstacles study. Small portions these be realistically simulated. Alone, smaller likely do not insight into collectively However, by proposing larger (containing many related copies system) have explicit connection, bottom-up CG techniques used transfer discovered using system primary interest. different prescribed (i) representation (mapping) (ii) functional form parameters represent energetics, which approximate potentials mean force (PMFs). design methods facilitate variety physically relevant representations, approximations, fields frontier forward. Crucially, parametrization interest orthogonal optimization potential present all methods. empirical efficacy machine learning tasks provides strong motivation consider approaches approximating PMF analyzing approximations.

Language: Английский

---

Artificial intelligence for natural product drug discovery

Nature Reviews Drug Discovery, Journal Year: 2023, Volume and Issue: 22(11), P. 895 - 916

Published: Sept. 11, 2023

Language: Английский

---

Application advances of deep learning methods for de novo drug design and molecular dynamics simulation

Wiley Interdisciplinary Reviews Computational Molecular Science, Journal Year: 2021, Volume and Issue: 12(3)

Published: Oct. 14, 2021

Abstract De novo drug design is a stationary way to build novel ligands in the confined pocket of receptor by assembling atoms or fragments, while molecular dynamics (MD) simulation dynamical study interaction mechanism between and receptors based on force field. MD are effective tools for discovery. With development technology, deep learning methods, interpretable machine (IML) have emerged research area design. Deep methods IML can be used further improve efficiency accuracy de simulations. The application summary design, simulations, promote technical In this article, two major workflow related components classical algorithm described from new perspective. progress also summarized Furthermore, introduced model interpretability Our paper deals with an interesting topic about applications simulations scientific community. This article categorized under: Data Science > Chemoinformatics Artificial Intelligence/Machine Learning

Language: Английский

---

AlphaFold, Artificial Intelligence (AI), and Allostery

The Journal of Physical Chemistry B, Journal Year: 2022, Volume and Issue: 126(34), P. 6372 - 6383

Published: Aug. 17, 2022

AlphaFold has burst into our lives. A powerful algorithm that underscores the strength of biological sequence data and artificial intelligence (AI). appended projects research directions. The database it been creating promises an untold number applications with vast potential impacts are still difficult to surmise. AI approaches can revolutionize personalized treatments usher in better-informed clinical trials. They promise make giant leaps toward reshaping revamping drug discovery strategies, selecting prioritizing combinations targets. Here, we briefly overview structural biology, including molecular dynamics simulations prediction microbiota-human protein-protein interactions. We highlight advancements accomplished by deep-learning-powered protein structure their impact on life sciences. At same time, does not resolve decades-long folding challenge, nor identify pathways. models provides do capture conformational mechanisms like frustration allostery, which rooted ensembles, controlled dynamic distributions. Allostery signaling properties populations. also generate ensembles intrinsically disordered proteins regions, instead describing them low probabilities. Since generates single ranked structures, rather than cannot elucidate allosteric activating driver hotspot mutations resistance. However, capturing key features, deep learning techniques use predicted conformation as basis for generating a diverse ensemble.

Language: Английский

---

Classical and reactive molecular dynamics: Principles and applications in combustion and energy systems

Progress in Energy and Combustion Science, Journal Year: 2023, Volume and Issue: 97, P. 101084 - 101084

Published: April 29, 2023

Molecular dynamics (MD) has evolved into a ubiquitous, versatile and powerful computational method for fundamental research in science branches such as biology, chemistry, biomedicine physics over the past 60 years. Powered by rapidly advanced supercomputing technologies recent decades, MD entered engineering domain first-principle predictive material properties, physicochemical processes, even design tool. Such developments have far-reaching consequences, are covered first time present paper, with focus on combustion energy systems encompassing topics like gas/liquid/solid fuel oxidation, pyrolysis, catalytic combustion, heterogeneous electrochemistry, nanoparticle synthesis, heat transfer, phase change, fluid mechanics. First, theoretical framework of methodology is described systemically, covering both classical reactive MD. The emphasis development force field (ReaxFF) MD, which enables chemical reactions to be simulated within framework, utilizing quantum chemistry calculations and/or experimental data training. Second, details numerical methods, boundary conditions, post-processing costs simulations provided. This followed critical review selected applications methods systems. It demonstrated that ReaxFF been successfully deployed gain insights pyrolysis oxidation fuels, revealing detailed changes pathways. Moreover, complex physico-chemical dynamic processes reactions, soot formation, flame synthesis nanoparticles made plainly visible from an atomistic perspective. Flow, transfer change phenomena also scrutinized simulations. Unprecedented nanoscale droplet collision, evaporation, CO2 capture storage under subcritical supercritical conditions examined at atomic level. Finally, outlook discussed context emerging computing platforms, machine learning multiscale modelling.

Language: Английский

---

A State-of-the-Art Review on Machine Learning-Based Multiscale Modeling, Simulation, Homogenization and Design of Materials

Archives of Computational Methods in Engineering, Journal Year: 2022, Volume and Issue: 30(1), P. 191 - 222

Published: Aug. 5, 2022

Language: Английский

---