Wiley Interdisciplinary Reviews: Computational Molecular Science最新文献

Integrating materials representations into feature engineering by rational design plays a critical role in determining the capability and accuracy of material property prediction via machine learning (ML). There still exists a lack of comprehensive classification and multi-dimensional evaluation for many existing feature models that could guide model selection in applications and further development. This review systematically classifies feature construction methods for crystalline structures, emphasizing the coupling between chemical and structural information. We systematically discuss the geometric configurations, chemical attributes, and their intricate coupling mechanisms that can be leveraged for feature engineering. Furthermore, a comprehensive comparison is performed across multiple aspects including graph network representation, structural information embedding, chemistry-structure information coupling, local versus global characteristics, long-range versus short-range description, algorithm compatibility with kernel function method or deep neural network, data size requirements, computational complexity, and interpretability mechanisms, thereby highlighting key variations in existing feature models and improving the physical interpretability of predictive models. To illustrate the integration of multi-dimensional characteristics, the center-environment (CE) feature model is introduced based on the coupling between local chemical and structural information of physical core-shell structures. Within the CE model, the pre-attention mechanism reorients focus from intricate details within complex ML algorithms to explicit feature models that depict physical core-shell configurations. By minimizing data requirements while enhancing transparency in ML models, the CE feature provides a practical approach for developing efficient and accurate ML-based predictions tailored for small-data scenarios in materials science.

This article is categorized under:

• Structure and Mechanism > Computational Materials Science
• Data Science > Artificial Intelligence/Machine Learning

Molecular polaritons are hybrid states formed by the quantum mechanical interaction between light and matter. Recent experiments have shown the ability to drastically modify chemical reactions in both the ground and excited states through the hybridization of the electronic and photonic degrees of freedom. Ab initio simulations of molecular polaritons have demonstrated similar effects for simple ground and excited state reactions. However, the theoretical community has been limited in its ability to describe the complicated dynamical processes of many-molecule collective effects with a high-level treatment of all degrees of freedom within a rigorous Hamiltonian. In this review, we provide a general description and overall procedure for exploring molecular polaritons, leveraging standard many-body electronic structure calculations combined with the exact, non-relativistic quantum electrodynamics light-matter Hamiltonian.

This article is categorized under:

• Electronic Structure Theory > Ab Initio Electronic Structure Methods
• Software > Quantum Chemistry
• Structure and Mechanism > Reaction Mechanisms and Catalysis

Computational bioprospecting is revolutionizing enzyme discovery by addressing key challenges associated with traditional laboratory and microbiological methods, such as resource-intensive experimentation and the limited cultivability of microorganisms. This review outlines current in silico methodologies, highlighting their effectiveness in identifying and prioritizing enzymes with desirable expression, stability, and catalytic activity properties. We emphasize recent advancements, including deep learning approaches and AlphaFold-based structure predictions, and discuss their integration with classical molecular mechanics techniques. Through our experiences—such as bioprospecting thermostable oxidases and high-activity laccases—we illustrate practical applications of machine learning, molecular simulations, and synthetic data generation to pinpoint promising enzyme candidates efficiently. Finally, we identify critical gaps, including data scarcity and the need for better integration of multi-omics information, which must be addressed to refine computational approaches in enzyme bioprospecting.

This article is categorized under:

• Structure and Mechanism > Computational Biochemistry and Biophysics
• Data Science > Artificial Intelligence/Machine Learning

DNA-based technologies for object authentication and data storage are becoming an interesting alternative to classic identification systems, yet their practical implementation faces fundamental technical and commercial barriers that limit widespread adoption. This review presents an analysis of DNA tagging and storage technologies, assessing their technical features, cost-effectiveness, and real-world applicability through comparison of competing approaches. We demonstrate that DNA tagging and data storage applications exhibit fundamentally different requirements, necessitating divergent technological strategies rather than unified solutions. DNA tagging faces severe cost disadvantages ($1–$100 per authentication versus $0.01–$0.10 for established technologies) and extended verification times (30 min to 6+ hours versus instant readout), limiting viability to high-security, low-volume markets such as pharmaceuticals and luxury goods. Current commercial implementations frequently lack peer-reviewed validation, creating an evidence deficit that undermines enterprise confidence. Among current approaches, isothermal amplification methods (LAMP, RPA) combined with colorimetric detection represent the most promising pathway for field-deployable authentication, while Illumina sequencing platforms provide optimal performance for data storage applications. The absence of standardization frameworks fundamentally constrains commercial adoption across both domains, preventing interoperability and enabling unsubstantiated performance claims. We conclude that successful commercialization requires strategic reorientation toward application-specific optimization and integrative approaches where DNA serves as secondary authentication combined with established identifiers, rather than competing directly on speed and cost metrics.

This article is categorized under:

• Structure and Mechanism > Molecular Structures
• Data Science > Databases and Expert Systems
• Molecular and Statistical Mechanics > Molecular Mechanics