Innovative strategies for the quantitative modeling of blood–brain barrier (BBB) permeability: harnessing the power of machine learning-based q-RASAR approach†
{"title":"Innovative strategies for the quantitative modeling of blood–brain barrier (BBB) permeability: harnessing the power of machine learning-based q-RASAR approach†","authors":"Vinay Kumar, Arkaprava Banerjee and Kunal Roy","doi":"10.1039/D4ME00056K","DOIUrl":null,"url":null,"abstract":"<p >In the current research, we have unveiled an advanced technique termed the quantitative read-across structure–activity relationship (q-RASAR) framework to harness the power of machine learning (ML) for significantly enhancing the precision of predictions related to blood–brain barrier (BBB) permeability. It is important to emphasize that the central objective of this study is not to introduce an additional model for predicting BBB permeability. Instead, our focus is on highlighting the improvement in quantitatively predicting the BBB permeability of organic compounds by utilizing the q-RASAR approach. This innovative methodology strives to enhance the precision of evaluating neuropharmacological implications and streamline the drug development process. In this investigation, we developed a machine learning (ML)-based q-RASAR PLS regression model using a large dataset comprising 1012 diverse classes of heterocyclic and aromatic compounds, obtained from the freely accessible B3DB database (accessible at https://github.com/theochem/B3DB) to predict BBB permeability during the lead discovery phase for central nervous system (CNS) drugs. The model's predictive capability underwent validation using two external sets, encompassing a total of 1 130 315 compounds, including synthetic compounds and natural products (NPs) for data gap filling and other two external sets comprising 116 drug-like/drug compounds from the FDA and ChEMBL databases to assess the model's reliability against the reported BBB permeability values. This study aimed to bridge the data gap by employing a predictive regression model to estimate the BBB permeability for both synthetic compounds and natural products (NPs). To further enhance predictability, we have developed various other ML-based q-RASAR models. The insights from the developed model highlight the pivotal roles played by hydrophobicity, electronic effects, degree of ionization, and steric factors as essential features facilitating the traversal of the blood–brain barrier. This research not only advances our understanding of the molecular determinants influencing the permeability of central nervous system drugs but also establishes a versatile computational platform for the rapid assessment of diverse compounds, facilitating informed decision-making in the realms of drug development and design.</p>","PeriodicalId":91,"journal":{"name":"Molecular Systems Design & Engineering","volume":" 7","pages":" 729-743"},"PeriodicalIF":3.2000,"publicationDate":"2024-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Molecular Systems Design & Engineering","FirstCategoryId":"5","ListUrlMain":"https://pubs.rsc.org/en/content/articlelanding/2024/me/d4me00056k","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CHEMISTRY, PHYSICAL","Score":null,"Total":0}
引用次数: 0
Abstract
In the current research, we have unveiled an advanced technique termed the quantitative read-across structure–activity relationship (q-RASAR) framework to harness the power of machine learning (ML) for significantly enhancing the precision of predictions related to blood–brain barrier (BBB) permeability. It is important to emphasize that the central objective of this study is not to introduce an additional model for predicting BBB permeability. Instead, our focus is on highlighting the improvement in quantitatively predicting the BBB permeability of organic compounds by utilizing the q-RASAR approach. This innovative methodology strives to enhance the precision of evaluating neuropharmacological implications and streamline the drug development process. In this investigation, we developed a machine learning (ML)-based q-RASAR PLS regression model using a large dataset comprising 1012 diverse classes of heterocyclic and aromatic compounds, obtained from the freely accessible B3DB database (accessible at https://github.com/theochem/B3DB) to predict BBB permeability during the lead discovery phase for central nervous system (CNS) drugs. The model's predictive capability underwent validation using two external sets, encompassing a total of 1 130 315 compounds, including synthetic compounds and natural products (NPs) for data gap filling and other two external sets comprising 116 drug-like/drug compounds from the FDA and ChEMBL databases to assess the model's reliability against the reported BBB permeability values. This study aimed to bridge the data gap by employing a predictive regression model to estimate the BBB permeability for both synthetic compounds and natural products (NPs). To further enhance predictability, we have developed various other ML-based q-RASAR models. The insights from the developed model highlight the pivotal roles played by hydrophobicity, electronic effects, degree of ionization, and steric factors as essential features facilitating the traversal of the blood–brain barrier. This research not only advances our understanding of the molecular determinants influencing the permeability of central nervous system drugs but also establishes a versatile computational platform for the rapid assessment of diverse compounds, facilitating informed decision-making in the realms of drug development and design.
期刊介绍:
Molecular Systems Design & Engineering provides a hub for cutting-edge research into how understanding of molecular properties, behaviour and interactions can be used to design and assemble better materials, systems, and processes to achieve specific functions. These may have applications of technological significance and help address global challenges.