By Dr Mahesh Bhalgat
Artificial intelligence (AI) presents the possibility to alter the drug discovery process and has grown significantly, in the past few years. Over 150 companies have applied AI-based drug discovery approaches to raise funds and progress molecules into clinical trials.
Before diving into the use of Artificial Intelligence in drug discovery, a very high-level approach to traditional drug discovery methods needs to be understood as it involves several elaborate, expensive, and time-consuming steps. Target identification involves identifying the right biological target, which may be a gene, protein, or transcript involved in a physiological pathway. This first step is followed by hit identification, which involves going through many levels of screening for generating lead compounds. Despite this extensive process of evaluating and optimizing the lead candidate, there are always uncertainties about a candidate’s progress to the next phase of development because of insufficient bioavailability, unacceptable toxicity, or the inability to replicate lab success in living systems.
There has been considerable interest in deciphering ways to improve the success rate in drug discovery, and AI and Machine Learning (ML) advances can help at every stage of the drug discovery process
The pharmaceutical industry generates hundreds of gigabytes of complex data of to identify the ideal small or large-molecule drug. High-throughput screening (HTS) analyzes thousands of chemical and biological compounds that researchers need to explore. These large datasets require appropriate analytical methods to yield statistically valid models and implementation of AI-based models leading to significant improvement in data utilization and smart decision-making.
This labour intensive and elaborate process also contributes to the cost of drugs. As reported, developing a drug from the discovery phase to its commercial manufacturing costs an average of $3 billion and takes 10-15 years. However, despite all efforts, the new candidates may fail in clinical trials and be unsuccessful in getting regulatory approvals. As per Eli Lilly’s data, only about 12 per cent of the potential drugs that start phase 1 trials end up in commercialization. The need of the hour is to design strategies in the early phase with the help of AI-based models to diminish the attrition of new drugs.
Key highlights of AI-driven drug discovery:
- AI can be implemented at various points of drug discovery to accelerate the process cut associated costs
- The availability of large datasets & the development of advanced algorithms have driven major improvements in ML
- AI driven iterative screening is a next-generation to- generation HTS
- Throughout the drug discovery value chain, automation has the potential to increase laboratory efficiency and engage talent in more complex tasks.
What AI-based technologies offer
AI techniques are increasingly needed to solve the complexity associated with managing chemistry, biology, toxicology and omics data sets; predict the toxicity and potency of the drug and empower scientists to understand complex human biology. Deep learning coupled with relevant modelling studies, assists in safety and efficacy evaluations of drug molecules based on big data modelling and analysis. These studies help to draw inferences from the available data with certainty, design drug discovery models and derive a conclusion from a hypothesis. Consequently, drug development costs are reduced and the speed of therapeutics development is expedited.
The efficacy of drug molecules depends on their affinity for the target and engagement to deliver the desired therapeutic response. Those with no to low interaction should be screened out along with those interacting with unwanted proteins or receptors, leading to toxicity. Computational methods and web-based tools can measure a drug’s binding affinity, predict drug toxicity to avoid lethal effects in subjects and reduce the number of animals used in experiments.
Current industry landscape and prospects
To ensure clinical success, one can develop a drug discovery platform that provides a deeper understanding of disease to identify and prioritize targets, based on, drug ability, and safety, thus decreasing the later-stage attrition. Recently, GlaxoSmithKline used Exscientia’s platform to identify a highly potent active lead molecule to treat chronic obstructive pulmonary disease. Generally, this involves making and testing hundreds or thousands of compounds over multiple iterative cycles. Exscientia produced 85 compounds in 5 cycles to obtain the lead candidate, showcasing the tremendous power of AI. Pfizer is using IBM Watson, a machine learning system, to power its search for immuno-oncology drugs. Roche is using an AI system from GNS Healthcare in Cambridge, to drive the hunt for cancer treatments. AstraZeneca used an AI platform to target idiopathic pulmonary fibrosis. At Syngene, the Syn.AITM platform helps researchers gain deeper data insights with advanced analytics to accelerate drug discovery programs.
The AI in the drug discovery market increased from US$200 million (2015) to US$700 million (2018) and is expected to increase to $5 billion by 2024. The first AI-based small-molecule drug candidates are now in clinical trials. Data-based technologies will improve the quality of products, ensure batch-to-batch consistency during production, and provide better utilization of available resources. Further, it will help develop the right therapy for patients (precision medicines for patients with cancer) and manage clinical trial data with a higher probability of success while characterizing patient subgroups that are most likely to benefit from the treatments.
Dr Mahesh Bhalgat COO, Syngene International.
(DISCLAIMER: The views expressed are sole of the author and ETHealthworld does not necessarily subscribe to it. ETHealthworld.com shall not be responsible for any damage caused to any person/organisation directly or indirectly.)