Cryo-electron microscopy (Cryo-EM) enables rapid freezing of samples, capturing biomolecules in a nearly natural state and providing highly accurate, undistorted images at near-atomic resolution. Image processing plays a crucial role in this process. Dr. Mingxu Hu has long been dedicated to developing theories and algorithms for Cryo-EM image processing, publishing several papers in top international journals such as Nature Methods, Nature Communications, Communications Biology, Journal of Structural Biology, Nature, Nature Microbiology, and PNAS.

He proposed the CryoSieve particle selection method, which significantly improved the resolution and efficiency of Cryo-EM, pushing its performance close to theoretical limits (Nature Communications, 2023). CryoPROS, another of his innovations, addresses particle orientation estimation challenges caused by preferred orientation in Cryo-EM by generating auxiliary particles through artificial intelligence (in review). CryoTRANS employs a self-supervised neural network to predict high-resolution conformations of rare conformations in Cryo-EM (Communications Biology, 2024).

Dr. Hu also introduced the use of quaternion representations for angles in Cryo-EM, providing a novel perspective for accurately describing and processing molecular structures (Journal of Structural Biology, 2020). Additionally, he developed a theoretical framework for particle orientation statistics under molecular symmetry and created the corresponding software package, pySymStat (in press). Furthermore, Dr. Hu was the first to propose and successfully implement precise per-particle correction of phase transfer function parameters, developing the software THUNDER, which significantly enhanced the accuracy of image processing (Nature Methods, 2018).

Dr. Hu’s current research interests include:

1.Developing high-throughput structural analysis techniques for Cryo-EM by integrating image data with amino acid sequence information using deep learning models.

2.Leveraging the symmetry of molecular scaffolds to significantly improve the signal-to-noise ratio in averaged structures, developing techniques for semi-in situ high-throughput Cryo-EM structural analysis in lysates, particularly chloroplast lysates.

3.Exploring the potential for directed evolution of biomolecules using high-throughput Cryo-EM structural analysis, especially the possibility of driving lipid synthesis using external energy sources.

Dr. Hu’s research group fosters an academic environment that is equal, open, and inclusive, and welcomes more researchers to join in advancing high-throughput Cryo-EM structural analysis and exploring the industrial applications of biomolecules.

The CryoSieve particle selection method developed by Mingxu Hu shows that high-resolution reconstruction can be achieved in cryo- electron microscopy (cryo- EM) using only a minimal number of particles near the theoretical limit, thus validating the theoretical feasibility of highthroughput cryo-EM (Nature Communications, 2023). Based on this approach, he further established a high- throughput cryo- EM pipeline for structural analysis of carbohydrate fibers under native conditions (LTS preprint server, 2025). Using this technique, he enabled large scale determination of carbohydrate fiber structures and built the CryoSeek database (https://cryoseek.org.cn), creating a key resource platform for glycobiology research.

In addition, he developed the Ahaha chirality determination method, which resolves a long standing challenge in chiral measurement for biological dark matter research (PNAS, 2026). For difficult datasets, his CryoPROS method uses artificial intelligence to generate auxiliary particles, effectively reducing orientation estimation bias caused by preferred particle orientation (Nature Communications, 2025). The CoCoFold algorithm fine- tunes AlphaFold predictions using highly limited cryo- EM data, bringing predicted structures into agreement with experimental observations (Communications Chemistry, 2026). CryoTRANS uses a self supervised neural network with velocity fields to enable high- resolution prediction of rare conformational states in cryo- EM (Communications Biology, 2024).

He also introduced quaternion based angular representation in cryo- EM, providing a new framework for the precise description and processing of molecular structures (Journal of Structural Biology, 2020). Furthermore, he developed a theoretical framework for particle orientation statistics under molecular symmetry and implemented the corresponding software package pySymStat (SIAM Journal on Imaging Sciences, 2025).

Dr. Mingxu Hu was the first to propose and successfully implement precise per- particle correction of phase transfer function parameters, accompanied by the development of the software THUNDER, which substantially improved the accuracy of cryo- EM image processing (Nature Methods, 2018).

2020     ShuiMu Scholar, Tsinghua University

2019     Excellent Research Award, School of Life Sciences, Tsinghua University

2018     Advanced Innovation Young Scientist, Advanced Innovation Center for Structural Biology, Tsinghua University

* for co-author; #for co-corresponding author.

1. Zhang, Q., Qin, L., Wang, T., Li, Z., Zhang, Y., Chen, S., Ge, Q., Wen, J., Yan, N. ^#, Wang, J. ^#, Hu, M. ^#, 2026. Absolute hand determination of glycofibrils from natural sources in cryo-EM. Proceedings of the National Academy of Sciences, 123 (9) e2531477123.

2. Liao J.*, Zheng D.*, Zhang H.*, Zhang L., Hu M.^#, Bao C. ^#, 2026. Fine-tune AlphaFold with limited cryo-EM observations. Communications Chemistry, 9, 95.

3. Hu, M.*,^#, Chen, S. *, Wang, T. *, Qin, L. *, Zhang, Q. *, Zhang, Y., Ge, Q., Chen, T., Li, M., Li, C., Xu, G., Gui, Q., Li, Z. ^#, Yan, N. ^#, 2025. CryoSeek identification of glycofibrils with diverse compositions and structural assemblies. LTS Preprint Server.

4. Zhang, H.*, Zheng, D.*, Wu, Q., Yan, N., Peng, H., Hu, Q., Peng, Y., Yan, Z., Shi, Z., Bao, C.^#, Hu, M.^#, 2025. CryoPROS: Correcting misalignment caused by preferred orientation using AI-generated auxiliary particles. Nature Communications, 16.

5. Zhang, Q.^*, Bao, C.^#, Lin, H.^#, Hu, M.^#, 2024. Averaging orientations with molecular symmetry in cryo-EM. SIAM Journal on Imaging Sciences, 17 (4), 2174-2195.

6. Fan, X.*, Zhang, Q.*, Zhang, H., Zhu, J., Ju, L., Shi, Z.^#, Hu, M.^#, Bao, C.^#, 2024. CryoTRANS: predicting high-resolution maps of rare conformations from self-supervised trajectories in cryo-EM. Communications Biology, 7 (1), 1058.

7. Cai, M.*, Zhu, J.*, Zhang, Q.*, Xu, Y., Shi, Z., Bao C.^#, Hu, M.^#, 2024. Enhancing Density Maps by Removing the Majority of Particles in Single Particle Cryogenic Electron Microscopy Final Stacks. Journal of Visualized Experiments, (207), e66617.

8. Zhu, J.*, Zhang, Q.*, Zhang, H., Shi, Z.^#, Hu, M.^#and Bao, C.^#, 2023. A minority of final stacks yields superior amplitude in single-particle cryo-EM. Nature Communications, 14(1), p.7822. (Top 25 Nature Communications physics articles of 2023; reported on the front page by China Science Daily)

9. Hu, M.^*, Zhang, Q.^*, Yang, J.^# and Li, X.^#, 2020. Unit quaternion description of spatial rotations in 3D electron cryo-microscopy. Journal of Structural Biology, 212(3), p.107601.

10. Hu, M.^*, Yu, H.^*, Gu, K.^*, Wang, Z., Ruan, H., Wang, K., Ren, S., Li, B., Gan, L., Xu, S., Yang, G.^#, Shen Y^#. and Li X.^#, 2018. A particle-filter framework for robust cryo-EM 3D reconstruction. Nature Methods, 15(12), pp.1083-1089.

https://smart.org.cn/career/research-lab/340.html