Publication：Crystal Growth & Design
Authors：GuangXu Sun, Xuetao Liu, Yuriy A. Abramov, Sten O. Nilsson Lill, Chao Chang, Virginia Burger, and Anders Broo
Date：March 1, 2022
We demonstrate the successful application of the state-of-the-art AstraZeneca in-house and XtalPi cloud-based virtual polymorph screening workflows in support of stable form selection of crystalline oxabispidine AZD1305, a pharmaceutical compound. Experimental solid form screening had found two polymorphic forms, A and B, with physical stabilities that appeared to be extremely close at ambient temperature. Such observation may make experimental and in silico support of the solid form selection a challenging task. Both computational approaches correctly predicted the ranking and geometry of the stable form B at 0 K. This level of information would be important and sufficient for project support at the late discovery stage. However, metastable form A was predicted by both workflows to be considerably less stable than form B, separated by multiple virtual forms in the lattice energy landscapes. In order to account for the experimentally observed close physical stabilities of forms A and B at ambient temperature, calculation of the free-energy landscape was performed using pseudo-supercritical path method. This allowed the demonstration that, while form B is significantly more stable at 0 K, the two forms display a very close physical stability at ambient temperature. The current work highlights the importance of using advanced virtual polymorph screening to get a more comprehensive insight into identifying the most stable form of a pharmaceutical compound under different experimental conditions.
Publication：Journal of Chemical Information and Modeling
Authors：Yuriy A. Abramov, GuangXu Sun, and Qun Zeng
Date：Feb. 28, 2022
Computational chemistry applications have become an integral part of the drug discovery workflow over the past 35 years. However, computational modeling in support of drug development has remained a relatively uncharted territory for a significant part of both academic and industrial communities. This review considers the computational modeling workflows for three key components of drug preclinical and clinical development, namely, process chemistry, analytical research and development, as well as drug product and formulation development. An overview of the computational support for each step of the respective workflows is presented. Additionally, in context of solid form design, special consideration is given to modern physics-based virtual screening methods. This covers rational approaches to polymorph, coformer, counterion, and solvent virtual screening in support of solid form selection and design.
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：Crystal Growth Design
Authors：Yuriy A. Abramov, Bochen Li, Chao Chang, Qun Zeng, GuangXu Sun, Gianpaolo Gobbo
Date：Aug. 30, 2021
The modern crystal structure prediction (CSP) technologies have proven to be accurate enough to provide a valuable support for a stable form selection in the pharmaceutical industry. We demonstrate that successful applications of the CSP predictions, in part, may be accounted for by favorable uncertainty distribution with the smallest absolute errors in the low relative crystal energy region. Such behavior is dictated by the lowest contribution of the systematic scaling error of dispersion-corrected density functional theory (DFT-D) approaches in this region. These considerations are validated by benchmarking studies of selected popular DFT-D approaches relative to post-Hartree–Fock (post-HF) calculations for representative molecular dimeric configurations in the virtual crystalline states of four pharmaceutical compounds. In addition, discussed are uncertainty distributions of DFT-D predictions of relative energies of eight ROY and five oxalyl dihydrazide (ODH) polymorphs relative to MP2D/HMBI and CCSD(T)/HMBI predictions, respectively.
Publication：Royal Society of Chemistry
Authors：Jiuchuang Yuan, Xuetao Liu, Simin Wang, Chao Chang, Qiao Zeng, Zhengtian Song, Yingdi Jin, Qun Zeng, Guangxu Sun, Shigang Ruan, Chandler Greenwell and Yuriy A. Abramov
Date：July 2, 2021
Cocrystals as a solid form technology for improving physicochemical properties have gained increasing popularity in the pharmaceutical, nutraceutical, and agrochemical industries. However, the list of potential coformers contains hundreds of molecules; far more than can be routinely screened and confirmed. Cocrystal screening experiments require significant amounts of active ingredients at an early project stage, and are expensive and time-consuming. Physics-based models and machine learning (ML) models have both been used to perform virtual cocrystal screening to guide experimental screening efforts, but both have certain limitations. Here, we present a combined ML/COSMO-RS fast virtual cocrystal screening method that proves to be significantly better than the sum of its parts in application to internal and external validation sets. To achieve that, we have defined the optimal threshold values of ML cocrystallization probability and COSMO-RS excess enthalpy of drug/coformer mixing for the combined coformer ranking. An approach to determine an applicability domain (AD) of the ML model has been implemented. The speed and accuracy of the new combined model allow it to be a good alternative to the physics-based CSP-based approach to support pharmaceutical projects with tight timeline and budget constraints.
Publication：Journal of Chemical Information and Modeling
Authors：Zhixiong Lin, Junjie Zou, Shuai Liu, Chunwang Peng, Zhipeng Li, Xiao Wan, Dong Fang, Jian Yin, Gianpaolo Gobbo, Yongpan Chen, Jian Ma, Shuhao Wen, Peiyu Zhang*, and Mingjun Yang*
Date：June 4, 2021
Free energy perturbation (FEP) has become widely used in drug discovery programs for binding affinity prediction between candidate compounds and their biological targets. However, limitations of FEP applications also exist, including, but not limited to, high cost, long waiting time, limited scalability, and breadth of application scenarios. To overcome these problems, we have developed XFEP, a scalable cloud computing platform for both relative and absolute free energy predictions using optimized simulation protocols. XFEP enables large-scale FEP calculations in a more efficient, scalable, and affordable way, for example, the evaluation of 5000 compounds can be performed in 1 week using 50–100 GPUs with a computing cost roughly equivalent to the cost for the synthesis of only one new compound. By combining these capabilities with artificial intelligence techniques for goal-directed molecule generation and evaluation, new opportunities can be explored for FEP applications in the drug discovery stages of hit identification, hit-to-lead, and lead optimization based not only on structure exploitation within the given chemical series but also including evaluation and comparison of completely unrelated molecules during structure exploration in a larger chemical space. XFEP provides the basis for scalable FEP applications to become more widely used in drug discovery projects and to speed up the drug discovery process from hit identification to preclinical candidate compound nomination.