DeepModeling

With the increasing demand for molecular representation learning in drug discovery and materials design, multimodal representation learning that combines the 3D geometric structure of molecules with biomedical text is becoming a research hotspot. Recently, a paper titled "GeomCLIP: Contrastive Geometry-Text Pre-training for Molecules" [1] published on arXiv introduced an innovative framework called GeomCLIP in detail. This framework combines the 3D geometric information of molecules with text descriptions and significantly enhances the multimodal learning ability of molecular representations through contrastive learning and denoising pre-training tasks, demonstrating superior performance in several downstream tasks.

The research team developed a large-scale dataset called PubChem3D, which contains more than 200,000 pairs of 3D geometry and text descriptions of molecules, covering rich chemical and biological information. The GeomCLIP framework adopts a dual-encoder structure to encode the geometry and text of molecules respectively and aligns the representations of the two modalities through contrastive learning while retaining the ability of the geometric encoder to model the characteristics of 3D molecules. Experiments show that GeomCLIP has achieved excellent performance in molecular property prediction, molecule-to-text retrieval, and 3D molecular caption generation tasks, providing important support for drug research and development and materials design.

This research was completed by a joint team from the Artificial Intelligence Research Laboratory at Pennsylvania State University and the Shenzhen International Graduate School of Tsinghua University. Teng Xiao from Pennsylvania State University is the first author and corresponding author of the paper, and the collaborators include Chao Cui, Huaisheng Zhu, and Professor Vasant G. Honava from Tsinghua University. This work has opened up a new direction for multimodal learning of molecular representations and also provided new solutions for drug discovery and materials science research.

Recently, the research group led by Zhiwei Li from the School of Physical Science and Technology at Lanzhou University and the research group led by Hanjie Guo from the Songshan Lake Materials Laboratory collaborated. A research paper titled "Short-range order and strong interplay between local and itinerant magnetism in GeFe₃N" published in "Physical Review B" used microscopic probing combined with DFT calculations (ABACUS+TB2J) to reveal the unique magnetic behavior of the GeFe₃N compound [1]. This compound belongs to the space group I4/mcm, has a complex crystal structure and unique microscopic magnetism, and has attracted extensive attention. This article will introduce the crystal structure, magnetic behavior, and the physical mechanism behind it of GeFe₃N based on this research paper. Master's student Tinghai Zhang is the first author, and Yantao Cao, Bo Zhang, and Liang Qiao provided experimental and data analysis support.

Water is not only one of the most familiar substances to humans but also a central figure in the long history of physical chemistry. The tetrahedral arrangement and network interactions between its molecules distinguish it from simple liquids.

For a long time, there has been no specific conclusion regarding whether there is a liquid - liquid critical point (LLCP) in water. Besides, researchers' understanding of water, especially when it acts as a solvent, is still incomplete.

To address the problem of technically and reasonably representing the thermodynamic and kinetic properties of water after the introduction of other chemical substances, a team from Seoul National University in South Korea proposed a scheme to examine the spatiotemporal characterization of water using a machine learning force field (MLFF) through deep potential molecular dynamics (DPMD).

This research, titled "Spatiotemporal characterization of water diffusion anomalies in saline solutions using machine learning force field", was published in "Science Advances" on December 11, 2024.

Currently, most water models are unable to fully capture the dynamic behavior of water after the addition of salt. Although classical force fields provide important insights, their simplifications and the omission of dynamic charge effects may distort our understanding of the real behavior of water.

The application advantages of MLFF in fields such as materials science and its processing speed, which is more than six orders of magnitude faster than first - principles methods in systems composed of several hundred atoms, make it stand out among all the options.

DeepModeling community has officially released Uni-Mol2, which is currently the largest 3D molecular representation foundation model. The largest version of Uni-Mol2 has a parameter scale of 1.1 billion and has been pre-trained on 800 million molecular conformations, demonstrating excellent performance in multiple molecular property prediction tasks. This achievement not only provides a powerful tool for deep learning research in the field of molecular science but also lays a solid experimental foundation for exploring larger-scale molecular pre-training models. At the NeurIPS 2024 conference currently being held in Vancouver, Canada, Uni-Mol2, as an accepted paper, has also received extensive attention.

Recently, Associate Professor Xu Yang from the School of Science at Shenyang Aerospace University, in cooperation with Professor Fei Du and Professor Yi Zeng from Jilin University and other scholars, conducted an in-depth study on the cubic phase K₃SbS₄ solid-state electrolyte based on the DeePMD method. The related research results were published in the journal "Chemistry of Materials" under the title "Cl-Doped Cubic K₃SbS₄ as a Solid-State Electrolyte for K-Ion Batteries with Ultrafast Ionic Conductivity" (DOI: 10.1021/acs.chemmater.4c02575).

On October 17, 2024, the research paper titled "Entropy in catalyst dynamics under confinement" by the AI4EC Lab/Professor Jun Cheng's research group from Xiamen University was published online in the international journal Chem. Sci. The first author of the paper is Qiyuan Fan(currently a teacher at the School of Chemistry and Chemical Engineering, Shanxi University). This work was completed under the guidance of Professor Jun Cheng and with the guidance and support of Academician Zhongqun Tian, Academician Xinhe Bao from the Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Academician Weinan E from Peking University, and Professor Ye Wang from Xiamen University.

GPUMD is an efficient domestic molecular dynamics simulation software developed and maintained by Professor Zheyong Fan from Bohai University. The software first released its public version 1.0 in 2017 [Computer Physics Communications 218, 10 (2017)] and has currently been iterated to version 3.9.4. GPUMD includes both commonly used empirical potentials and NEP (Neuroevolution Potential) machine learning potentials. Up to now, GPUMD has been used by thousands of users in many countries around the world and has attracted dozens of researchers to participate in its development. It is widely applied in fields such as heat and mass transfer, mechanical properties, structural phase transitions, irradiation damage, spectroscopy, and catalysis. Related achievements have been published in top academic journals such as Nature, Nature Communications, J. Am. Chem. Soc, ACS Nano, Phys. Rev. Lett, J. Mech. Phys. Solids, J. Chem. Theory Comput., Phys. Rev. B, and J. Chem. Phys.

In June 2024, GPUMD&NEP joined the DeepModeling community. As an innovative and highly efficient MD simulation and machine learning potential function tool, it further provides support for the Materials Genome Project and the AI4S community.

With the growing market demand for efficient and safe rechargeable batteries that can operate under extreme temperature conditions, the rapid and accurate evaluation of key properties of electrolyte molecules has become particularly important. A recent paper titled "A Knowledge–Data Dual-Driven Framework for Predicting the Molecular Properties of Rechargeable Battery Electrolytes," published in Angewandte Chemie International Edition, details an innovative approach known as the "Knowledge–Data Dual-Driven Framework" (KPI) specifically designed to predict the molecular properties of battery electrolytes, including melting point (MP), boiling point (BP), and flash point (FP). The research team skillfully combined deep learning techniques with domain-specific chemical knowledge, supported by large-scale datasets, significantly enhancing the accuracy and efficiency of predictions. In this framework, the Uni-Mol model plays a central role, demonstrating great potential in predicting the properties of electrolyte molecules and providing strong support for the development of next-generation high-performance batteries.

Xiang Chen, an associate research fellow in the Department of Chemical Engineering at Tsinghua University, is the corresponding author of the paper. Yuchen Gao, a 2022 direct-entry PhD student in the Department of Chemical Engineering, is the first author. Co-authors include Yuhang Yuan, an undergraduate from the Tsinghua Academy of Wisdom; Suozhi Huang, an undergraduate from the Institute for Interdisciplinary Information Sciences; Nan Yao, a 2020 direct-entry PhD student; Legeng Yu, a 2021 direct-entry PhD student; Yaopeng Chen, a 2023 direct-entry PhD student; and Qiang Zhang, a professor in the Department of Chemical Engineering. The research was supported by funding from the National Natural Science Foundation of China, the National Key R&D Program of China, and the Beijing Natural Science Foundation.

On August 19, 2024, Linlin Hou, Hongxin Xiang, and Xiangxiang Zeng from Hunan University, in collaboration with Li Zeng from the Shanghai Institute of Materia Medica, published a research article titled "Attribute-guided Prototype Network for Few-shot Molecular Property Prediction" in Briefings in Bioinformatics. This study introduced an Attribute-guided Prototype Network (APN), which innovatively combines high-level molecular fingerprints with deep learning algorithms, significantly improving the accuracy of molecular property prediction in limited-sample scenarios. This breakthrough opens a new direction for few-shot learning in drug development.

The third highlight of DeePMD-kit v3 is its model plugin mechanism. This article introduces the feature from three perspectives: background and principles, development tutorial, and usage tutorial.