Po-Jen Hsu

Work Experience

• 2016/8 - present: Postdoctoral Research Fellow, Institute of Atomic and Molecular Sciences, Academia Sinica.

• 2015/8 - 2016/7: Postdoctoral Research Fellow, Department of Applied Chemistry, National Chiao Tung University.

• 2015 - 2015/7: Senior Data Scientist, Vpon Big Data Mobile Advertising -- Improved advertisement traffic quality and built real-time bidding engine.

• 2014 - 2015: Senior Engineer, Innovation Digital System, System Software Development Division, Hon Hai Precision IND. CO., LTD. -- Built hand gesture control algorithms based on computer vision for IoT environment and supported camera subsystem in Android cell phone.

Education

• • 2014: Ph.D., Molecular Science and Technology, TIGP, Academia Sinica/Physics, National Central University

  • 2003: M.S. in Physics, National Central University

  • 2000: B.S. in Physics, National Central University

Research Interests

In parallel to developing advanced experimental techniques with better resolution to understand the structures in the experiment, it is also highly desirable to have the capability to interpret the experimental results independently from theoretical approaches. In fact, many measurable characteristics in the experiment usually reveal the presence of many isomeric structures or conformers. In the past few years, we have developed a systematic structural analysis based on the first-principle calculation for various complex molecular systems. A flexible framework for constructing structure database was proposed to deal with large amount of structure samples. With the structure database, we are able to investigate the molecular system in many aspects including the dissociation process, the assignment of IR spectrum, and the temperature dependence of structure population. Our analysis highlights the fact that diversity in the intrinsic molecular structures lays the foundations of different characteristics of a chemical or biological system

Hydrogen bond network structures of protonated clusters

    Hydrogen-bonded (H-bonded) clusters in the gas phase are an ideal model system to explore the H-bonded network structures. It has also been demonstrated that the preferential H-bonded network structure strongly depends on the species of the protonated cluster. For example, the H-bonded networks of protonated water clusters develop from radial tree structures to three-dimensional cage structures as the size increases. On the other hand, protonated methanol clusters form much simpler (bi)cyclic or linear chain structure even in large sizes. The origin of such differences is obviously the topological property of the H-bonded networks in different cluster isomers. Previously, we had performed extensive structure search on protonated methanol, ethanol, and tert-butanol clusters. The temperature-dependent population of isomers and the simulated IR spectra are computed through the quantum harmonic superposition approximation (Q-HSA) of the local minimum.

    Our isomer structure search plays an important role to mimic different types of H-bond networks presented in the experiment. Here we do not focus on the global structure solely since we realized that hydrogen bonded system can be highly energy degenerated. We turned to analyze different H-bond topologies among the isomers we collected using genetic algorithm and other sampling techniques. To efficiently screen out duplicates, we also developed a sophisticated structure clustering algorithm associated with pattern recognition techniques for the H-bonded networks and cluster shapes. Our algorithm makes sure the resulting isomers are all structurally distinct. Finally, we apply the density functional theory to accurately depict the vibrational frequencies of the clusters. Our previous attempts on the smaller and simpler H-bond system show good agreement with the experimental IR spectra measured by Prof. Asuka Fujii's laboratory in Tohoku university , encourage us to understand the more complex H-bond system such as mixed alcohol clusters.

Identification of polysaccharide structures through computational structure sampling

Our structure searching algorithm is feasible to the structural identification of saccharides, which plays an important role in carbohydrate analysis. The main interest in this study is to treat the dihedral angle of two attached hexose isomers (mono-saccharides with 6 carbon atoms) as a free parameter and compute the simulated vibrational spectra with respect to different types of composed hexose isomers and different dihedral angles. This approach allows us to mimic the condition of the disaccharide samples in the experiment and helps to explain the difficult question of resolving the IR or collision-induced dissociation (CID) spectra of the polysaccharides. This work will also provide important structural database of the disaccharide molecules for researchers in Prof. Chi-Kung Ni's lab to interpret the experimental CID results, and for researchers in Prof. Jer-Lai Kuo's lab to perform high-level anharmonic frequency calculations.

Development of structural searching for complex molecules

• 1.  We apply molecular dynamics and Monte Carlo simulations to search the conformations of sugar and hydrogen bonded molecular systems.

  2. We adopt SQL database to store thousands or more than 10 thousands molecular structures and use Pandas python module to carry out Big Data mining and analysis.

  3. We use Django python module to construct websites for molecular structure database associated with D3.js technology for data visualization.

  4. We aim to develop an automatic and efficient solution for structural searching in a molecular system. Our work will be published on a web server or integrated as a Open-source software.

Selected Papers

• Competition between hydrogen bonds and van der Waals force in intermolecular structure formation of protonated branched-chain alcohol clusters, Natsuko Sugawara, Po-Jen Hsu, Asuka Fujii, and Jer-Lai Kuo, Phys. Chem. Chem. Phys. (accepted)

  • Collision-Induced Dissociation of Sodiated Glucose, Galactose, Mannose, and the Identification of Anomeric Configuration, Hai Thi Huynh, Huu Trong Phan, Po-Jen Hsu, Jien-Lian Chen, Hock-Seng Nguan, Shang-Ting Tsai, Chia-Yen Liew, Chi-Kung Ni, and Jer-Lai Kuo, Phys. Chem. Chem. Phys. 20, 19614 (2018)

  • Hydrogen Bond Network Structures of Protonated Short-chain Alcohol Clusters, Asuka Fujii, Natsuko Sugawara, Po-Jen Hsu, Takuto Shimamori, Ying-Cheng Li, Toru Hamashima, and Jer-Lai Kuo, Phys. Chem. Chem. Phys. (Perspective) 20, 14971-14991 (2018)

  • Temperature and Size Dependence of Characteristic Hydrogen-Bonded Network Structures with Ion Core Switching in Protonated (Methanol)6-10 -(Water)1 Mixed Clusters: A Revisit, Marusu Katada, Po-Jen, Hsu, Asuka Fujii, Jer-Lai Kuo, J. Phys. Chem. A 121, 5399 (2017)

• • Exploration of hydrogen bond networks and potential energy surfaces of methanol clusters with a two-stage clustering algorithm, P. J. Hsu, K. L. Ho, S. H. Lin, and J. L. Kuo, Phys. Chem. Chem. Phys. 19, 544 (2017)

  • Precursory Signatures of Protein Folding/Unfolding: From Time Series Correlation Analysis to Atomistic Mechanisms, P. J. Hsu, S. A. Cheong, and S. K. Lai, J. Chem. Phys. 140, 204905 (2014).

  • Peptide dynamics by molecular dynamics and diffusion theory methods with improved basis sets, P. J. Hsu, S. K. Lai, and A. Rapallo, J. Chem. Phys. 140, 104910 (2014).

  • Dynamical study of metallic clusters using the statistical method of time series clustering, S. K. Lai, Y. T. Lin, P. J. Hsu, and S. A. Cheong, Compt. Phys. Commun. 182, 1013 (2011).

  • Melting scenario in metallic clusters, P. J. Hsu, J. S. Luo, S. K. Lai, J. F. Wax, and J-L Bretonnet, J. Chem. Phys. 129, 194302 (2008).

  • Structure of bimetallic clusters, P. J. Hsu and S. K. Lai, J. Chem. Phys. 124, 044711 (2006).

  • Structures of metallic clusters: mono- and polyvalent metals, S. K. Lai, P. J. Hsu, K. L. Wu, W. K. Liu, and M. Iwamatsu, J. Chem. Phys. 117, 10715 (2002).