08. Debate

The Great Chemistry Debate

2024 Event Descriptions

Coordinators: Dr. Patrick DePaolo and Prof. Alexei Khalizov

The Use of Machine Learning in Drug Discovery

Artificial intelligence (AI) has been a hot topic for several years. Many aspects of our lives are governed by AI and machine learning (ML), including social media and electronics. AI is also utilized in many areas of pharmaceutical drug discovery. Scientists have been attempting to integrate AI techniques into the traditional drug discovery workflow to increase hit to lead efficiency which would save pharmaceutical companies millions of dollars. The general drug discovery process is outlined below. AI integration has a potential to speed up the hit to lead process and can predict chemical and biological drug traits so they don’t all have to be tested in the lab. Often times machine learning models are used to test millions of small molecules against a drug target to assess binding likelihood. This large pool of compounds is narrowed down to a few hundred compounds that can be individually tested. This process adds an element of automation to the drug discovery workflow and increases efficiency, in theory.

There have been many challenges in the development of AI-driven protocols in drug discovery. Although the purpose is to speed up the process of finding clinical drug candidates, the industry may need significant improvement to these protocols. In 2010, a time where AI/ML methods were low, every $1 invested in drug discovery yielded a 10 cent return on average. This means that most pharmaceutical investors do not get a return on their investment. In 2020, many companies have implemented AI/ML into their drug discovery workflow with little success and only a 2 cent return for every $1 invested in drug discovery. This, along with the abysmal 90% failure rate of all drug discovery programs may show that AI is not the right direction for the pharmaceutical industry. The industry needs to see better molecule success predictions that can be used to accelerate drug development.

The question: Should we use AI/ML methods in modern drug discovery?

Consider the following:

  1. What is AI/ML? How does it work?
  2. How does drug discovery work?
  3. What are AI/ML methods in drug discovery and how can they theoretically accelerate R&D?
  4. What are the limitations of some of these methods?
  5. Does the promise of AI/ML methods outweigh the risk of lower efficiency and return on investment?

Part 1 – Conduct general literature research on the issue:

  1. Prepare a five-page essay with citations (Times New Roman font, double-spaced, 12 pt.  The essay must conclude with an advocacy position answering the question of whether AI/ML techniques should be used in drug discovery.  Students must also submit two questions, one for a Pro position opponent and one for a Con position opponent, to ask the opposing team on Event Day.

 

  1. Prepare oral presentations on each side of the debate (YES, AI/ML techniques should be used in drug discovery, and NO, AI/ML techniques should not be used in modern drug discovery).  Students must be prepared to argue either side on Event Day.

 

  1. Optional:  Prepare ONE slide for each side of the debate (YES, NO) that can be displayed during your oral presentation.

 

  1. REQUIRED:  Submit your essay and the two questions for opposing sides (YES, NO) by APRIL 17 (before Event Day!) for the judges to review beforehand and prepare questions on.

 

The essay should address:

  • What is AI/ML and how is it used in pharmaceuticals?
  • Several techniques used at different stages of drug development
  • The risks of heavily relying on these techniques to drive drug development
  • The rewards of successful integration in the drug development workflow.

Each team:

Will be randomly assigned either the YES or NO position by the judges.

 

 

Each team must present their assigned position (YES or NO) on the question in a 3 - 5 minute oral presentation, optionally with the ONE slide they prepared on that position

The Judges will ask each team one question about this initial presentation.   Each team will take two minutes to address their question from the Judges.

Each team will ask another team one question (previously submitted by their team) to be answered in one minute, after two minutes of deliberation by the answering team.

Each team must have each member speak at least once during any of the aspects (initial or response to questions asked) of the presentation.  The initial presentation can be done by one member or divided up into sections for different team members, not to exceed the total time as above.

 

The oral presentation and the submitted essay will be judged pursuant to the following rubric (1 to 5 points for each category):

Clarity and organization

Reasoning and creativity

Use of supporting facts

Reference material

Persuasiveness

Note that providing (brief) examples of alternative emerging techniques that can accelerate drug discovery may also be useful.