Prescription drugs have the ability to transform a patient’s life and provides them the opportunity to rid a devastating illness, making the development and approval of these medications urgent and necessary. However, over the last 10 years the path leading to drug approval has become more complicated and expensive. The cost to develop a new prescription medicine that gains market approval has gone up 145% to $2.6 billion and takes an average of 10 years to develop1. For patients suffering from an illness with no approved treatment, the wait can be unnerving. To this day, more than 400 million people suffer from rare diseases and 95% of rare diseases lack an FDA approved treatment3. Artificial Intelligence (AI), including its subfields of machine learning (ML) and deep learning (DL), have the power to change these statistics for the better.
AI can reduce the amount of time it takes to receive an accurate diagnosis and streamline the path to discover and develop treatments for devastating illnesses. This is just one therapeutic area where AI can have a large impact. If we think about research in more traditional areas such as oncology, gastroenterology, and CNS disorders where large amounts of data have already been produced, AI, ML, and DL can help discover new compounds that could become new drugs, uncover or repurpose drugs, and improve the area of personalized medicine.
How is AI being applied in Drug Discovery efforts?
It is reported that every one of the major pharmaceutical companies has announced a partnership with at least one AI-based drug development company, but how is AI being used to facilitate drug development? As mentioned, the cost to develop a new pharmaceutical treatment has increased, as well as the timeline to get a new treatment to the market. In part, this increase is due to the complexity of the disorders being treated. The drugs used to treat the most common disorders have already been found. The focus in the industry has now shifted, and researchers are searching for treatments to address complex rare diseases that involve hundreds of proteins, and singly targeting just one of those proteins is no longer enough.
The drug discovery process begins with the identification of a target for a drug to treat. For complex diseases, this requires that researchers fully understand how molecules will interact with one another. This is where AI comes into play. Companies such as Cyclica have developed software that matches the biophysical and biochemical properties of millions of molecules to the structures and properties of approximately 150,000 proteins to uncover molecules that are likely to bind to a target2. Merck and Bayer are among the big pharma companies that have announced a partnership with Cyclica to aid in their drug discovery efforts. While many pharma companies will not release much information regarding exactly what AI-generated drug candidates are coming out of their collaborations, Cyclica has shared some details of its successes in identifying a key target protein that is already linked to an FDA-approved drug.
However, there are instances where a researcher identifies a target protein but is unable to determine a molecule that will bind with the target protein. Companies such as Celgene, GSK, Sanofi, and Sunovion have partnered with Exscientia to solve this problem. Exscientia’s algorithm compares information available about a target protein against a database of about a billion interactions. By performing this step, the list of possible compounds that may work is narrowed, and details of what additional data would help to further refine the list of possible compounds is provided. This process is repeated until a manageable list of favorable drug compounds is generated2. GSK has reported that its partnership with Exscientia has led to a promising molecule targeting a novel pathway to treat chronic obstructive pulmonary disease4. The CEO of Exscientia, Andrew Hopkins, claims that their specialized process can reduce the time spent in drug discovery from 4.5 years to as little as one year, and reduces discovery costs by 80 percent.
Centralizing Data Management to Facilitate Machine Learning
Exscientia and Cyclica are just two examples of companies using AI to improve the drug discovery process and get life transforming treatments into the hands of patients more quickly. A key piece of facilitating AI is ensuring that all research data is readily available in a centralized location. Thermo Fisher Scientific is at the forefront for enabling centralized data management for laboratories. Thermo Fisher™ Platform for Science™ software enables centralized data management of laboratory data. Thermo Scientific™ Core SDMS™ and Core Connect™ software work with Platform for Science software to provide direct communication between laboratory instruments and software systems. By utilizing these products, all data can be stored in one central system, facilitating machine learning. Visit our website if you are interested in learning more about how Platform for Science software can centralize your laboratory data.
- Duggan, Susan, and Norma Pence. “A Tough Road: Cost To Develop One New Drug Is $2.6 Billion; Approval Rate for Drugs Entering Clinical Development Is Less Than 12%.” Policy & Medicine, 21 Mar. 2019, www.policymed.com/2014/12/a-tough-road-cost-to-develop-one-new-drug-is-26-billion-approval-rate-for-drugs-entering-clinical-de.html.
- Freedman, David H. “Hunting for New Drugs with AI.” Nature, vol. 576, no. 7787, 2019, doi:10.1038/d41586-019-03846-0.
- “RARE Facts.” Global Genes, globalgenes.org/rare-facts/.
- Smith, Simon. “36 Pharma Companies Using Artificial Intelligence in Drug Discovery.” 36 Pharma Companies Using Artificial Intelligence in Drug Discovery, 11 Dec. 2019, blog.benchsci.com/pharma-companies-using-artificial-intelligence-in-drug-discovery.