Explainable chemical artificial intelligence from accurate machine learning of real-space chemical descriptors

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: May 21, 2024

Abstract Machine-learned computational chemistry has led to a paradoxical situation in which molecular properties can be accurately predicted, but they are difficult interpret. Explainable AI (XAI) tools used analyze complex models, highly dependent on the technique and origin of reference data. Alternatively, interpretable real-space employed directly, often expensive compute. To address this dilemma between explainability accuracy, we developed SchNet4AIM, SchNet-based architecture capable dealing with local one-body (atomic) two-body (interatomic) descriptors. The performance SchNet4AIM is tested by predicting wide collection quantities ranging from atomic charges delocalization indices pairwise interaction energies. accuracy speed breaks bottleneck that prevented use chemical descriptors systems. We show group indices, arising our physically rigorous atomistic predictions, provide reliable indicators supramolecular binding events, thus contributing development Chemical Artificial Intelligence (XCAI) models.

Language: Английский

---

Grid: A Python library for molecular integration, interpolation, differentiation, and more

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 160(17)

Published: May 3, 2024

Grid is a free and open-source Python library for constructing numerical grids to integrate, interpolate, differentiate functions (e.g., molecular properties), with strong emphasis on facilitating these operations in computational chemistry conceptual density functional theory. Although designed, maintained, released as stand-alone library, was originally developed integration, interpolation, solving the Poisson equation HORTON ChemTools packages. designed be easy use, extend, maintain; this why we use adopt many principles of modern software development, including comprehensive documentation, extensive testing, continuous integration/delivery protocols, package management. We leverage popular scientific packages, such NumPy SciPy, ensure high efficiency optimized performance grid development. This article official release note showcasing its unique functionality scope.

Language: Английский

---

A rumen-derived bifunctional glucanase/mannanase uncanonically releases oligosaccharides with a high degree of polymerization preferentially from branched substrates

Carbohydrate Polymers, Journal Year: 2024, Volume and Issue: 330, P. 121828 - 121828

Published: Jan. 18, 2024

Language: Английский

---

Current developments and trends in quantum crystallography

Acta Crystallographica Section B Structural Science Crystal Engineering and Materials, Journal Year: 2024, Volume and Issue: 80(4), P. 249 - 274

Published: June 18, 2024

Quantum crystallography is an emerging research field of science that has its origin in the early days quantum physics and modern when it was almost immediately envisaged X-ray radiation could be somehow exploited to determine electron distribution atoms molecules. Today can seen as a composite area at intersection crystallography, chemistry, solid-state physics, applied mathematics computer science, with goal investigating problems, phenomena features crystalline state. In this article, state-of-the-art will described by presenting developments applications novel techniques have been introduced last 15 years. The focus on advances framework multipole model strategies, wavefunction-/density matrix-based approaches chemical topological techniques. Finally, possible future improvements expansions discussed, also considering new experimental computational technologies.

Language: Английский

---

Formulating Reduced Density Gradient Approaches for Noncovalent Interactions

The Journal of Physical Chemistry A, Journal Year: 2024, Volume and Issue: 128(30), P. 6158 - 6166

Published: July 23, 2024

This work elucidates several forms of reduced electron density gradient (RDG) to describe noncovalent interactions (NCIs). By interpreting the RDG as a local moment function, we systematically leveraged Weizacker's and Fermi's moments. resulted in high-fidelity representations consistent with NCI analysis. In addition, version derived from Lagrangian kinetic energy is conveniently normalized. These results suggest nonexistence particular formulation when performing Thus, an in-depth examination theoretical foundations connecting function nature necessary.

Language: Английский

---

CuGBasis: High-performance CUDA/Python library for efficient computation of quantum chemistry density-based descriptors for larger systems

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(7)

Published: Aug. 19, 2024

CuGBasis is a free and open-source CUDA®/Python library for efficient computation of scalar, vector, matrix quantities crucial the post-processing electronic structure calculations. integrates high-performance Graphical Processing Unit (GPU) computing with ease flexibility Python programming, making it compatible vast ecosystem libraries. We showcase its utility as demonstrate seamless interoperability existing software to gain chemical insight from quantum chemistry Leveraging GPU-accelerated code, cuGBasis exhibits remarkable performance, highly applicable larger systems or large databases. Our benchmarks reveal 100-fold performance compared alternative packages, including serial/multi-threaded Central GPU implementations. This paper outlines various features computational strategies that lead cuGBasis’s enhanced guiding developers code.

Language: Английский

---

Pentafluoroorthotellurate Uncovered: Theoretical Perspectives on an Extremely Electronegative Group

Inorganic Chemistry, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 3, 2025

The pentafluoroorthotellurate group (−OTeF5, teflate) exhibits high electron-withdrawing properties. Indeed, it is often used as a bulky substitute for fluoride due to its chemical stability and larger size, which reduces tendency act bridging ligand. These characteristics make valuable ligand in synthetic chemistry, facilitating the preparation of molecular structures analogous polymeric fluoride-based compounds. In this study, we explore electronic structure teflate by using advanced Quantum Chemical Topology (QCT) methods better understand bonding nature compare electronegativity with that halogens. For that, examine XOTeF5 systems (X = F, Cl, Br, I) decompose X–OTeF5 interactions into classical (ionic) exchange-correlation (covalent) contributions interacting quantum atoms (IQA) energy decomposition scheme. We also conduct detailed analysis electron distribution utilizing statistical framework functions (EDFs) localization function (ELF), density, reduced density gradient scalar functions, well delocalization indices QTAIM charges. results show properties are comparable those fluorine, albeit slightly lower. Moreover, internal primarily ionic. Additionally, −OTeF5 other O-donor groups, demonstrating within OEF5 (E S, Se, Te) nearly identical, these groups higher than OCF3, OC(CF3)3, OC6F5.

Language: Английский

---

Aromaticity monitoring during the donor-acceptor coordination of pyrosilicate, pyrophosphate, and pyrosulfate anions to a new designed anion receptor

Phosphorus, sulfur, and silicon and the related elements, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 10

Published: April 16, 2025

Language: Английский

---

Questing for Homoleptic Mononuclear Manganese Complexes with Monodentate O-Donor Ligands

Chemical Science, Journal Year: 2024, Volume and Issue: 15(15), P. 5564 - 5572

Published: Jan. 1, 2024

Compounds containing Mn-O bonds are of utmost importance in biological systems and catalytic processes. Nevertheless, mononuclear manganese complexes all O-donor ligands still rare. Taking advantage the low tendency pentafluoroorthotellurate ligand (teflate, OTeF

Language: Английский

---

Critical assessment of the x-ray restrained wave function approach: Advantages, drawbacks, and perspectives for density functional theory and periodic ab initio calculations

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 160(23)

Published: June 20, 2024

The x-ray restrained wave function (XRW) method is a quantum crystallographic technique to extract functions compatible with experimental diffraction data. approach looks for that minimize the energies of investigated systems and also reproduce sets structure factors. Given strict relationship between factors electron distributions, strategy practically allows determining correspond given (usually experimental) densities. In this work, capabilities XRW were further tested. aim was evaluate whether could serve as tool suggesting new exchange-correlation functionals density functional theory or refining existing ones. Additionally, ability address influences crystalline environment assessed. outcomes computations thus compared those traditional gas-phase, embedding mechanics/molecular mechanics, fully periodic calculations. results revealed that, irrespective initial conditions, yield consensus density, in contrast currently employed approximations (DFAs), which tend give too large range distributions. This encouraging view exploiting develop improved functionals. Conversely, calculations emphasized limited its effectively environment. underscores need technique, would allow untangling shortcomings DFAs from inherent approach.

Language: Английский

---