Authors：Qiaohao Liang, Lipeng Lai
Date：Dec. 13, 2021
While Bayesian Optimization (BO) has emerged as sample-efficient optimization method for accelerating drug discovery, it has rarely been applied to the process optimization of pharmaceutical manufacturing, which traditionally has relied on human-intuition, along with trial-and-error and slow cycles of learning. The combinatorial and hierarchical complexity of such process control also introduce challenges related to high-dimensional design spaces and requirements of larger scale observations, in which BO has typically scaled poorly. In this paper, we use penicillin production as a case study to demonstrate the efficacy of BO in accelerating the optimization of typical pharmaceutical manufacturing processes. To overcome the challenges raised by high dimensionality, we apply a trust region BO approach (TuRBO) for global optimization of penicillin yield and empirically show that it outperforms other BO and random baselines. We also extend the study by leveraging BO in the context of multi-objective optimization, allowing us to further evaluate the trade-offs between penicillin yield, production time, and CO2 emission as by-product. Through quantifying the performance of BO across high-dimensional and multi-objective drug production optimization processes, we hope to popularize application of BO in this field, and encourage closer collaboration between machine learning and broader scientific communities.
Publication：Journal of Physical Chemistry B
Authors：Muhammad A. Hagras, Michael A. Bellucci, Gianpaolo Gobbo, Ryan A. Marek, and Bernhardt L. Trout*
Date：Oct. 28, 2020
Disulfide cross-linking is one of the fundamental covalent bonds that exist prevalently in many biological molecules that is involved in versatile functional activities such as antibody stability, viral assembly, and protein folding. Additionally, it is a crucial factor in various industrial applications. Therefore, a fundamental understanding of its reaction mechanism would help gain insight into its different functional activities. Computational simulation of the disulfide cross-linking reaction with hydrogen peroxide (H2O2) was performed at the integrated quantum mechanical/molecular mechanical (QM/MM) level of theory in a water box under periodic boundary conditions. A benchmarking study for the barrier height of the disulfide formation step was performed on a model system between methanethiol and methane sulfenic acid to determine, for the QM system, the best-fit density functional theory (DFT) functional/basis set combination that produces comparable results to a higher-level theory of the coupled-cluster method. Computational results show that the disulfide cross-linking reaction with H2O2 reagent can proceed through a one-step or a two-step pathway for the high pKa cysteines or two different pathways for the low pKa cysteines to ultimately produce the sulfenic acid/sulfenate intermediate complex. Subsequently, those intermediates react with another neutral/anionic cysteine residue to form the cysteine product. In addition, the solvent-assisted proton-exchange/proton-transfer effects were examined on the energetic barriers for the different transition states, and the molecular contributions of the chemically involved water molecules were studied in detail.
Authors：Xue-Xiang Zhang, Huan Qi, Ya-Lan Liu, Song-Qiu Yang, Peng Li, Yan Qiao, Pei-Yu Zhang, Shu-Hao Wen, Hai-long Piao and Ke-Li Han
Date：Sept. 21, 2020
The applications of most fluorescent probes available for Glutathione S-Transferases (GSTs), including NI3 which we developed recently based on 1,8-naphthalimide (NI), are limited by their short emission wavelengths due to insufficient penetration. To realize imaging at a deeper depth, near-infrared (NIR) fluorescent probes are required. Here we report for the first time the designing of NIR fluorescent probes for GSTs by employing the NIR fluorophore HCy which possesses a higher brightness, hydrophilicity and electron-deficiency relative to NI. Intriguingly, with the same receptor unit, the HCy-based probe is always more reactive towards glutathione than the NI-based one, regardless of the specific chemical structure of the receptor unit. This was proved to result from the higher electron-deficiency of HCy instead of its higher hydrophilicity based on a comprehensive analysis. Further, with caging of the autofluorescence being crucial and more difficult to achieve via photoinduced electron transfer (PET) for a NIR probe, the quenching mechanism of HCy-based probes was proved to be PET for the first time with femtosecond transient absorption and theoretical calculations. Thus, HCy2 and HCy9, which employ receptor units less reactive than the one adopted in NI3, turned out to be the most appropriate NIR probes with high-sensitivity and little nonenzymatic background noise. They were then successfully applied to detecting GST in cells, tissues and tumor xenografts in vivo. Additionally, unlike HCy2 with a broad isoenzyme selectivity, HCy9 is specific for GSTA1-1, which is attributed to its lower reactivity and the higher effectiveness of GSTA1-1 in stabilizing the active intermediate via H-bonds based on docking simulations.
Publication：Crystal Growth Design
Authors：Mingjun Yang, Eric Dybeck, Guangxu Sun, Chunwang Peng, Brian Samas, Virginia M. Burger, Qun Zeng, Yingdi Jin, Michael A. Bellucci, Yang Liu, Peiyu Zhang, Jian Ma, Yide Alan Jiang, Bruno C. Hancock, Shuhao Wen*, and Geoffrey P. F. Wood
Date：Sept. 2, 2020
Crystal structure prediction (CSP) calculations can reduce risk and improve efficiency during drug development. Traditionally, CSP calculations use lattice energies computed through density functional theory. While this approach is often successful in predicting the low energy structures, it neglects the crucial role of thermal effects on polymorph stabilities. In the present study, we develop a robust and efficient protocol for predicting the relative stability of polymorphs at different temperatures. The protocol is executed on a highly parallel cloud computing infrastructure to produce results at time scales useful for drug development timelines. We demonstrate this protocol on molecule XXIII from the sixth crystal structure prediction blind test. Our results predict that Form D is the most stable experimentally observed polymorph at ambient temperature and Form C is the most stable at low temperature consistent with experiments also conducted in the present study.
Publication：Journal of Chemical Information and Modeling
Authors：Junjie Zou, Jian Yin, Lei Fang, Mingjun Yang, Tianyuan Wang, Weikun Wu, Michael A. Bellucci, and Peiyu Zhang*
Date：July 28, 2020
The ability of coronaviruses to infect humans is invariably associated with their binding strengths to human receptor proteins. Both SARS-CoV-2, initially named 2019-nCoV, and SARS-CoV were reported to utilize angiotensin-converting enzyme 2 (ACE2) as an entry receptor in human cells. To better understand the interplay between SARS-CoV-2 and ACE2, we performed computational alanine scanning mutagenesis on the “hotspot” residues at protein–protein interfaces using relative free energy calculations. Our data suggest that the mutations in SARS-CoV-2 lead to a greater binding affinity relative to SARS-CoV. In addition, our free energy calculations provide insight into the infectious ability of viruses on a physical basis and also provide useful information for the design of antiviral drugs.
Publication：Organic Process Research & Development
Authors：Yu-hong Lam*, Yuriy Abramov, Ravi S. Ananthula, Jennifer M. Elward, Lori R. Hilden, Sten O. Nilsson Lill, Per-Ola Norrby, Antonio Ramirez, Edward C. Sherer, Jason Mustakis, and Gerald J. Tanoury*
Date：July 23, 2020
Application of computational methods to understanding and predicting properties of analogues for drug discovery has enjoyed a long history of success. However, the drug development space (post-candidate selection) is currently experiencing a rapid growth in this arena. Due to the revolution in computing hardware development and improved computational techniques, quantum chemical (QC) calculations have become an essential tool in this space, allowing results from complex calculations to inform chemical development efforts. As a result, numerous pharmaceutical companies are employing QC as part of their drug development workflow. Calculations cover the range of transition state calculations, reaction path determination, and potential energy surface scans, among others. The impact of this rapid growth is realized by providing an in-depth understanding of chemical processes and predictive insight into the outcome of potential process routes and conditions. This review surveys the state of the art in these drug development applications in the pharmaceutical industry. Statistics of computational methods, software, and other metrics for publications in the last 14 years (2005–2019) are presented. Predictive applications of quantum chemistry for influencing experiments in reaction optimization and catalyst design are described. Important gaps in hardware and software capabilities that need to be addressed in order for quantum chemistry to become a more practical and impactful tool in pharmaceutical drug development are discussed. Perspectives for the future direction of application of QC to pharmaceutical drug development are proposed.