Machine Learning Assisted Molecular Simulations - Mass Spectrometry Data Analysis and Rational Design of Antibacterial Co3O4 Nanowires Flagella - Doctor of Philosophy Dissertation

Aug 2, 2026

Upload by Patrick

This dissertation presents machine learning–assisted computational workflows that connect molecular simulation, mass spectrometry data analysis, and rational nanomaterial design to advance antibacterial research. It develops an implicit solvent approach for aqueous–solution molecular dynamics and quantum-mechanics modeling, and introduces electrostatics-driven ensemble learning for pKa prediction. A transferable meta-learning framework is proposed for single-cell mass spectrometry phenotype prediction with limited melanoma samples. Finally, it reports a bioengineered bacterial flagella templating route to synthesize polycrystalline Co3O4 nanowires for gram-negative antibacterial applications.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages97

Reading Time4 h 4 min

AccessFree to download

1 Machine learning based implicit solvent model for aqueous–solution Ala nine dipeptide molecular dynamics simulations

1.1 Introduction

1.2 Machine Learning Based Implicit Solvent Model for MM and QM Modeling of Solute Molecules

1.3 Computational Details

1.4 Results and discussion

1.5 Conclusions

2 Electrostatics-Driven ensemble learning approach for pKa predictions

2.1 Introduction

2.2 Methodology

2.3 Results and Discussion

2.4 Conclusions

3 MetaPhenotype: A transferable meta learning model for single-cell mass spectrometry based cell phenotype prediction using a limited sample of melanoma cells

3.1 Abstract

3.2 Introduction

3.3 Methods

3.4 Result and Discussion

3.5 Conclusion

4 Bioengineered bacterial flagella-templated in situ green synthesis of polycrystalline Co3 O4 nanowires for gram-Negative antibacterial applications

4.1 Abstract

4.2 Introduction

4.3 Materials and Methods

UNIVERSITY OF OKLAHOMA GRADUATE COLLEGE MACHINE LEARNING ASSISTED MOLECULAR SIMULATIONS / MASS SPECTROMETRY DATA ANALYSIS AND RATIONAL DESIGN OF ANTIBACTERIAL Co3 O4 NANOWIRES FLAGELLA A DISSERTATION SUBMITTED TO THE GRADUATE FACULTY in partial fulfillment of the requirements for the Degree of DOCTOR OF PHILOSOPHY By SONGYUAN YAO Norman, Oklahoma MACHINE LEARNING ASSISTED MOLECULAR SIMULATIONS / MASS SPECTROMETRY DATA ANALYSIS AND RATIONAL DESIGN OF ANTIBACTERIAL Co3 O4 NANOWIRES FLAGELLA A DISSERTATION APPROVED FOR THE CHEMISTRY AND BIOCHEMISTRY DEPARTMENT BY THE COMMITTEE CONSISTING OF Dr. Yihan Shao, Chair Dr. Bin Wang Dr. Si Wu Dr. Bayram Saparov © Copyright by SONGYUAN YAO 2023 All Right Reserved. iv Acknowledgements I am deeply appreciative to everyone who contributed to my PhD thesis. The journey to this point has been filled with challenges, but the support and encouragement of numerous people in my life has made this possible. First, I would want to express my deepest gratitude to Dr. Yihan Shao, my PhD advisor. Throughout my PhD program, he has served as my guide, mentor, and advocate. Iam grateful for his continuous encouragement and support, as well as his crucial assistance in defining my research work. His skills and commitment to my studies have been an inspiration and a source of motivation. I am also grateful to the members of my committee members, Dr. Bin Wang, Dr. Si Wu and Dr. Bayram Saparov. Their insightful suggestions and constructive criticism have assisted me in enhancing the quality of my work. Their experience in their different professions has helped shape my studies and extend my perspective. My appreciation also extends to the following colleagues: Dr. Xiaoliang Pan, Richard Van, Chance Lander, Carly Wickizer, Eric Calderon Leon, Tra D. Nguyen and Yunpeng Lan. Their assistance, cooperation, and friendship have enhanced my study endeavors. I also would like to thank all the individuals I met at OU, Dr. Lin Wang, Dr. MengMeng Zhai, Dr. Penghe Qiu, and Dr. Binrui Cao. Everyone of them has taught me so much, and their contributions to my work have been essential. I am extremely grateful to my parents who unconditionly support me in my whole life and my fiance, Dr. Wen Yang for her unwavering understanding throughout the past seven years. Meanwhile, Dr. David Chissoe from the OU counseling center has been my mentor and was assisting me through the difficult period of time. Their support and trust in me have maintained me throughout the difficult times. Their love and constant support have been the foundation of my achievement, and I will be truly thankful for that. Finally, I would want to express my appreciation to everyone who has helped me throughout my academic path. This thesis is the product of years of work, and it would not have been possible without the support and encouragement of the numerous people who have been a part of this journey. I will always be grateful for the support and encouragement I have gotten, and I will always be thankful for this experience. v Contents 1 Machine learning based implicit solvent model for aqueous–solution ala nine dipeptide molecular dynamics simulations 1 1.1 Introduction .................................... 1 1.2 Machine Learning Based Implicit Solvent Model for MM and QM Modeling of Solute Molecules ................................ 6 1.2.1 Descriptor for the solute molecule .................... 7 1.2.2 The fitting neural networks (FNN) and loss function ......... 8 1.3 Computational Details .............................. 9 1.3.1 Training/validation configurations ................... 10 1.3.2 Labelling data .............................. 11 1.3.3 Training of the machine-learning models ................ 12 1.3.4 Umbrella Sampling ............................ 13 1.4 Results and discussion .............................. 14 1.4.1 MLP implicit solvent model for MM modeling .............

FAQs

What are the core research themes of this dissertation?

The dissertation integrates machine learning with molecular simulations and mass spectrometry analysis, then applies the outcomes to rational antibacterial material design using Co3O4 nanowires templated by bacterial flagella.

How is implicit solvent modeling addressed in the work?

It proposes a machine learning based implicit solvent model for aqueous-solution simulations, including both MM and QM modeling of solute molecules, with specific descriptor design and neural-network fitting.

What machine learning approaches are used for pKa prediction and single-cell mass spectrometry?

For pKa prediction, it uses an electrostatics-driven ensemble learning approach. For single-cell mass spectrometry, it introduces MetaPhenotype, a transferable meta-learning model that works with limited melanoma samples.

Aug 2, 2026

Upload by Patrick