Discovering New Drugs

Do you want to live longer? Think about a heart condition, stroke, tumors, cancer.

Do you want to have a better life? Think about depression, Alzheimer's, overweight, blood pressure, stomach ulcer, menopause.

A few researchers in Finland are working on drug discovery: searching for new molecules which could be used for medicinal purposes. Making new drugs is expensive, so most of the research in Finland focuses on the early phases of the process. However, there are also groups which handle the making of the molecules (synthesis), experimentation, and clinical testing.

I'm especially interested in computational tools and methods in drug discovery. Using computers for drug discovery is an appealing idea, but in practice it is difficult to implement a reliable procedure. Most of the bottlenecks are caused by lack of knowledge and tools. So you have to be satisfied with approximations and sometimes even with intuition.

This is an interesting time. If the Finnish researchers at some point have a breakthrough molecule, it will take a dozen years and perhaps hundreds of millions of euros to develop an effective medicine.

My background is in physics and mathematics. I have only recently begun to appreciate the difficulties in understanding how a human being functions, both on the molecular and on the cellular level. To make it possible to use modeling and computing in drug discovery, you have to make all kinds of shortcuts.

How computational drug discovery works? First, you model the structure and function of the target protein. Then you search the molecular databases for drug candidates, or design new molecules by yourself.

For each candidate molecule you evaluate the pharmaceutical properties, using perhaps custom-made tools. When you finally have a set of candidate molecules, you have to evaluate their potential as drugs. And if your partners and the funding agencies are satisfied, you may move into building the molecules (synthesis) and making experiments.

All this takes expertise, time, and money. So there is a pressing need for new methods and approaches for making more reliable and more comprehensive searches for molecules.

My expertise on optimization seems to be of use in drug discovery. I have worked with traditional gradient-based optimization methods. I have also studied the new methods, such as simulated annealing and genetic algorithms. It is nice to see that advanced optimization methods are used even in commercial software for drug discovery. Of course, to make these methods work you have to carefully tune them to the problems.

There is a pressing need for a computational platform in drug discovery: databases, modeling, and computing. First, you have to develop interfaces for exchanging data between different applications. Then you have to develop easy but flexible user interfaces, hoping to automate the drug discovery process for the researchers. The platform could be based on standard (and portable) tools, such as Java, Perl, PHP, and XML-based formats. You should use standards and tools developed elsewhere, and combine these with tools developed for your own projects.

The exchange, storage, and retrieval of data is a central problem in drug discovery. You have to handle structural information of the molecules, chemical and pharmaceutical properties, and results of computations, and perhaps integrate all this with biological or medicinal data.

The central problem for all academic institutions in drug discovery is the access to information. The big drug companies can purchase the expensive databases and user interfaces, but the situation is different for small startups and academic researchers. I hope there will be good tools available free of charge to academic researchers.