China 
Africa 
Explore chapters and articles related to this topic
AI and Drug Discovery
Published in Sandeep Reddy, Artificial Intelligence, 2020
Arash Keshavarzi Arshadi, Milad Salem
One important branch of CADD is virtual screening (VS). VS is a process to predict the potency of the compounds in inhibition of micro-organism or a molecule-based target like an enzyme (Rollinger, Stuppner, and Langer 2008). Target-based (rational) VS and non-target-based (irrational) VS are two main fields of VS (Lionta et al. 2014; Ripphausen, Nisius, and Bajorath 2011). Each of Target-based approach includes at least two main subsections: structural-based virtual screening (SBVS) and ligand-based virtual screening (LBVS) (Ripphausen et al. 2011). SBVS consists of any type of computer-aided prediction for compounds’ activity through inhibiting a molecular target with known 3D structure. Molecular docking is a popular example of SBVS (Huang, Shoichet, and Irwin 2006). It mostly consists of determining the binding affinity of the target and the ligand. The target would be any kind of biomolecule such as lipid, DNA and protein. The model can work with the simulation of interactions or complementary surface. On the other hand, LBVS mostly predicts compounds’ potency minimally by understanding the molecular patterns in the hits with mostly considering no information about the cell of interest. Cluster analysis (CA) would be a good example of the LBVS. Determining the similarity of the compounds is the basis of CA (Abramyan et al. 2016).
Molecular Aspects of the Activity and Inhibition of the FAD-Containing Monoamine Oxidases
Published in Peter Grunwald, Pharmaceutical Biocatalysis, 2019
By far the most important computational advance for drug discovery has been the ability to search databases of millions of compounds for “hits”—compounds with a pharmacophore that will have good affinity for a target-binding site. With a clinically and biologically validated target identified, crystallized proteins and proven ligands enable the development of the descriptors for key interactions with the target binding site. The pharmacophore derived from effective, selective molecules allows chemi-informatics to identify similarity-based compounds and, with advanced data-mining, to exclude binding to undesirable off-targets (Nikolic et al., 2015). The dual resource of ligand and target information provides the basis for virtual screening that is now routine in drug discovery. Objective ranking of the hits can then be achieved by docking and model refinement. Molecular dynamics to check the ligand to protein fit is also important, allowing for the target flexibility, and subsequent optimization of the ligand structure. A much shorter list must then be synthesized and experimentally verified. The process has recently been summarized with examples across target fields (Ramsay et al., 2018).
Artificial Intelligence in Systems Biology
Published in P. Kaliraj, T. Devi, Artificial Intelligence Theory, Models, and Applications, 2021
S. Dhivya, S. Hari Priya, R. Sathishkumar
In pharmacogenomics, AI technology plays a vital role in converting the accumulated raw data into a readable format. At present, pharmaceutical and nutraceutical industries have started using systems biology and machine learning algorithms to determine the molecular mechanism of the drug thereby reducing the time and cost of the experiment. The development of a novel drug is generally based on clinical studies, electronic medical imaging, medical records, and DNA expression profiling. In hospitals, a huge volume of clinical reports can be analyzed via machine learning algorithms. A case study involves the implementation of AI in systems biology towards pharmacogenomics. For instance, the application of computational tools and machine learning techniques in pharmacogenomics helps in the discovery of a precise drug for cancer treatment with the help of patient details, drug resistance, NGS data, and gene expression profiling. Accurate and efficient drug molecules are being developed by a virtual screening approach. Machine-learning methods and AI technology have been functionalized in diverse stages of virtual screening. There are two stages of virtual screening: (1) ligand-structure based virtual screening and(2) ligand-receptor based binding. Potential AI algorithm generates ligand-structure based virtual screening methods, which are based on non-parametric scoring functions. To identify the exact ligand site, various non-predetermined AI-based scoring functions are utilized (RF-score, ID score, and NN score). Intending to improve the performance of scoring functions in AI technology uses four algorithms: (1) SVM, (2) Bayesian, (3) RF, and (4) DNN methods. Nagarajan et al. (2019) detected the mutants and identified a precise drug target using a computational approach (Figure 7.14).
AIDrugApp: artificial intelligence-based Web-App for virtual screening of inhibitors against SARS-COV-2
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2023
Coronavirus Disease 2019 (COVID-19) is a current pandemic respiratory infection caused by a positive-sense RNA virus called severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) reported in December 2019 (Sohrabi et al., 2020). Several Covid-19 therapeutic agents have been evaluated but there has been no evidence of antiviral efficacy (Nsanzabaganwa et al., 2020). The largest research has been performed to date on Remdesivir (Shrestha et al., 2021), hydroxychloroquine (Dragojevic Simic et al., 2021) and Lopinavir (Stower, 2020), Ritonavir (Hariyanto et al., 2020) which poses concerns about its efficacy as a COVID-19 treatment. Deep learning (DL) is a major contributor to biological science research and drug discovery. Previous studies have suggested that deep neural network techniques have demonstrated superior performance to other machine learning algorithms in virtual screening, which is a key step in accelerating drug development (Zhang, Tan, Han et al., 2017; Carpenter et al., 2018; Korotcov et al., 2017). While the powerful performance of deep learning technologies in drug discovery is widely recognised in both academia and industry, adequate user-friendly tools and interfaces are still limited (Naudé, 2020). Therefore, we have developed an easy-to-use web app for virtual molecular screening to a specific target of interest for SARS-CoV-2. This web-app is based on deep neural network (DNN) models with supervised machine learning algorithms. It helps to predict inhibitory activities and pIC-50 (microM) values of new compounds towards Replicase polyprotein (RP; Kumar et al., 2020), Angiotensin Converting Enzyme (ACE; Chatterjee & Thakur, 2020), 3CLpro (Rathnayake et al., 2020) and molecules related to Clinical trials (CT) for SARS-COVID-19 diseases.
Pharmacoinformatics-based strategy in designing and profiling of some Pyrazole analogues as novel hepatitis C virus inhibitors with pharmacokinetic analysis
Published in Egyptian Journal of Basic and Applied Sciences, 2023
Stephen Ejeh, Adamu Uzairu, Gideon A. Shallangwa, Stephen E. Abechi, Muhammad Tukur Ibrahim
Because of ADMET risks, many prospective medications never make it to the clinic. Due to “their importance, ADMET qualities are now being examined in early-stage pharmaceutical research, resulting in a significant reduction in the number of molecules that failed in clinical trials due to poor ADMET properties [26,30]. Computational virtual screening of selected compounds (17), designed molecules, and the reference inhibitor was used to assess drug-likeness, including oral bioavailability and synthetic accessibility, utilizing Lipinski’s Ro5 (Table 3).