Streamlining Drug Discovery through Open Innovation
A wonderful example of how tapping smart people who don't work for you can deliver a solution you would never have dreamed of.
Computer scientists develop a novel technique to help the development of safe and effective medicines.
Merck & Co., United States
Bringing a new drug to market is a long, costly and expensive. The traditional industry approaches involve testing hundreds of thousands of compounds against potential disease mechanisms. Despite the technology coming on leaps and bounds in recent years, this was described as "an expensive, difficult, and inefficient process" in 'Identifying Cardiotoxic Compounds', an article that appeared in Genetic
Engineering & Biotechnology News in 2009.
Working with smart external sources of knowledge is one way some pharma companies are attempting to solve research problems and speed up timescales. In several situations the crowd has equipped itself well in solving challenging dilemmas that have vexed pharma industry experts.
Innovating through Data Analysis
To try and crack a particularly tough problem, that of streamlining its drug discovery process, pharmaceutical company Merck enlisted the help of Kaggle, a predictive analytics crowdsourcing site. It wanted a quicker and more efficient method of identifying molecules that are effective against intended targets but do not trouble non-intended targets and cause side effects.
So, Merck and Kaggle created a crowdsourcing contest (The Merck Molecular Activity Challenge) to find the best statistical techniques for predicting the biological activities of molecules of interest, both on and off target.
Merck released 15 data sets on chemical compounds it had previously tested, and invited participants to identify those that held the most promise for future testing. The contest ran for 60 days and attracted more than 2,000 proposals from 238 teams.
The winning solution didn't come from a pharmaceutical industry professional, but a team helmed by a computer scientist from the University of Toronto. Their winning solution, for which they were awarded $22,000 used neural networks and deep learning algorithms.
"After his previous experiences applying deep learning techniques to speech recognition and language processing tasks, George was drawn to the complexity of the Merck data set and the challenge of working in a new data domain,” said a post on Kaggle’s blog site. “He assembled a team of heavy hitters from the world of machine learning and neural networks.”
Potential Innovation Partners are Everywhere
The winning team came up with a methodology unfamiliar to Merck, which the pharma giant may not have come across had it not opened up the challenge to the crowd. It is a neat illustration of how an answer that you didn't even know existed can come from somewhere you might not even look.
Next Story »