Modern chemical graph theory

Wiley Interdisciplinary Reviews Computational Molecular Science, Год журнала: 2024, Номер 14(5)

Опубликована: Сен. 1, 2024

Abstract Graph theory has a long history in chemistry. Yet as the breadth and variety of chemical data is rapidly changing, so too do graph encoding methods analyses that yield qualitative quantitative insights. Using illustrative cases within basic mathematical framework, we showcase modern theory's utility Chemists' analysis model development toolkit. The both experimental simulation discussed at various levels granularity information. This followed by discussion two major classes theoretical analyses: identifying connectivity patterns partitioning methods. Measures, metrics, descriptors, topological indices are then introduced with an emphasis upon enhancing interpretability incorporation into physical models. Challenging described include strategies for studying time dependence. Throughout, incorporate recent advancements computer science applied mathematics propelling new domains study. article categorized under: Molecular Statistical Mechanics > Dynamics Monte‐Carlo Methods Structure Mechanism Computational Materials Science Structures

Язык: Английский

---

Graph-Based Deep Learning Models for Thermodynamic Property Prediction: The Interplay between Target Definition, Data Distribution, Featurization, and Model Architecture

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown

Опубликована: Янв. 9, 2025

In this contribution, we examine the interplay between target definition, data distribution, featurization approaches, and model architectures on graph-based deep learning models for thermodynamic property prediction. Through consideration of five curated sets, exhibiting diversity in elemental composition, multiplicity, charge state, size, impact each these factors accuracy. We observe that i.e., using formation instead atomization energy/enthalpy, is a decisive factor, so careful selection approach. Our attempts at directly modifying result more modest, though not negligible, accuracy gains. Remarkably, molecule-level predictions tend to outperform atom-level increment predictions, contrast previous findings. Overall, work paves way toward development robust with universal capabilities, can reach excellent across sets compound domains.

Язык: Английский

---

Momentum gradient-based untargeted poisoning attack on hypergraph neural networks

Neurocomputing, Год журнала: 2025, Номер unknown, С. 129835 - 129835

Опубликована: Март 1, 2025

Язык: Английский

---

Introduction to Machine Learning for Predictive Modeling of Organic Materials

Challenges and advances in computational chemistry and physics, Год журнала: 2025, Номер unknown, С. 43 - 60

Опубликована: Янв. 1, 2025

Язык: Английский

---

A multi-modal transformer for predicting global minimum adsorption energy

Nature Communications, Год журнала: 2025, Номер 16(1)

Опубликована: Апрель 4, 2025

Abstract The fast assessment of the global minimum adsorption energy (GMAE) between catalyst surfaces and adsorbates is crucial for large-scale screening. However, multiple sites numerous possible configurations each surface/adsorbate combination make it prohibitively expensive to calculate GMAE through density functional theory (DFT). Thus, we designed a multi-modal transformer called AdsMT rapidly predict based on surface graphs adsorbate feature vectors without site-binding information. model effectively captures intricate relationships atoms cross-attention mechanism, hence avoiding enumeration configurations. Three diverse benchmark datasets were introduced, providing foundation further research challenging prediction task. Our framework demonstrates excellent performance by adopting tailored graph encoder transfer learning, achieving mean absolute errors 0.09, 0.14, 0.39 eV, respectively. Beyond prediction, AdsMT’s scores showcase interpretable potential identify most energetically favorable sites. Additionally, uncertainty quantification was integrated into our models enhance trustworthiness predictions.

Язык: Английский

---

Comment on “Molecular hypergraph neural networks” [J. Chem. Phys. 160, 144307 (2024)]

The Journal of Chemical Physics, Год журнала: 2024, Номер 161(20)

Опубликована: Ноя. 27, 2024

Язык: Английский

---

Multi-granularity physicochemical-inspired molecular representation learning for property prediction

Expert Systems with Applications, Год журнала: 2024, Номер unknown, С. 126115 - 126115

Опубликована: Дек. 1, 2024

Язык: Английский

---