Many drugs turned out to be accidental discovery – this is a very astonishing fact in the field of drug discovery. But to find a substance that can treat a specific disease among the countless substances in nature, hard work is very necessary in either biomarker screening or other steps of drug discovery. In most cases, scientists are expanding the screening target as indefinite as possible in order to find the target molecule. People usually use high-throughput screening (HTS) for hit to lead screening. HTS is very expensive. It is automatically completed by robots and aims to find the compounds with the greatest potential to reach the target and increase the chance of hit identification. At present, the industry is trying to use artificial intelligence (AI) to develop virtual screening technology to replace or enhance the traditional high-throughput screening (HTS) process and increase the speed and success rate of screening.

So, how can AI be used in this field?

Actually, AI can be applied to different aspects of drug development, including new drug development, drug effectiveness/safety prediction, construction of new drug molecules, screening of biomarkers, and research of new combination therapies.

Prediction of effectiveness and safety of new drugs
To turn a substance into a drug, it must have the characteristics of safety and effectiveness, and predicting in advance in the development of new drugs can greatly improve the success rate of R&D. By combining drug mining with AI, and by analyzing databases through IBM supercomputers and using deep learning neural networks, the structure-activity relationships of compounds can be analyzed so as to assess the risks of new drugs early in drug development.

In addition, deep learning network can also be developed to identify the basic modules in medicinal chemistry, affinity prediction and toxicity testing.

Candidate drug mining
It is understood that drug mining is the earliest and fastest-growing field of AI application. Computer simulations can predict drug activity, safety, and side effects. In this computer-aided drug design (CADD) process, data analysis is an initial key step, which understands and analyzes a large number of biological scientific information, including patents, genomic data and more than 10,000 publications uploaded daily by all biomedical journals and databases, through deep learning and natural language processing. Next, deep learning software ingests and analyzes information, finds associations and proposes corresponding drug candidates, and further screens molecular structures that are effective for certain diseases, such as molecules that can be used for neurodegenerative diseases but without heart or liver side effects.

Designing new drug molecules
Scientists usually design drug molecules by studying structure-activity relationships or studying the structure of target biological macromolecules. But with AI, things could be different. Instead of searching for lead compounds through trial and error, researchers can tell generative adversarial networks (GANs), a newer deep learning technology, to produce such compound molecules. GANs use two competitive neural network models to create new data that is different from the real data. Some companies have applied them to cancer drug development. In brief, this method of developing and training new molecular structures using Generative Adversarial Networks (GANs) can greatly reduce the time and costs of finding substances with potential drug properties.

Screening for biomarkers
Biomarker refers to biochemical indicators that can mark changes or possible changes in the structure or function of systems, organs, tissues, cells, and subcellular cells. In the medical field, biomarkers can be used to diagnose diseases (for example, prostate-specific antigen PSA is a biomarker for prostate cancer diagnosis), to judge the stage of disease, or to evaluate the safety and effectiveness of new drugs or new therapies in target populations.

AI technology can also be of help in finding new biological indicators and biomarkers for early cancer via screening up to 25 thousands of samples. Furthermore, AI can also help to conduct new drug research and development to find invasive breast cancer treatment solutions that do not respond to existing drugs. The screening process of AI platform is as follows:

New drug targets and combination therapies
A drug target refers to the binding site between a drug and a biological macromolecule. The selection of a novel and effective drug target is the primary task of new drug development.

Cloud-based artificial intelligence platform could also be used in the development of new anticancer drugs. First, researchers use this tool to analyze a large number of isolated data sources, including licensed and publicly available data, and to verify hypotheses. Real-time interaction yields evidenced results for the discovery of new drug targets in immunotherapy, and for research of combination therapies as well as for the selection of patient treatment strategies.

Author's Bio: 

BOC Sciences is a chemical vendor, supplying various inhibitors, metabolites, impurities and other reagents for research use.