Computer Systems Laboratory Colloquium

4:15PM, Wednesday, April 3, 2002
NEC Auditorium, Gates Computer Science Building B03

Creating Open-Source Tools for Drug Discovery
How Free Software Might Save Your Life

Warren L. DeLano, Ph.D.
Sunesis Pharmaceuticals/DeLano Scientific
About the talk:

Biology and medicine are in the midst of a broad revolution resulting from advances including the human genome project, DNA chips, protoemics, automated synthesis, and high-throughput screening technologies. Research is becoming profoundly information-driven, and innately dependent on computational software. However, many important software tools for analyzing biological and chemical information are not available to most scientists because of traditional, restrictive approaches to software development and distribution.

In order for biomedical science to progress at an optimal pace, we must adopt a new paradigm for creation and dissemination of core computational technologies. Fortunately in the past decade, while chemists and biologists were busy developing new experimental technologies, computer scientists invented essential infrastructure to support a new approach for biomedical software. Specifically, collaborative internet development tools and modular dynamically-linked programming languages make open-source development a realistic and superior means by which scientists can create interoperable tools and share them in true academic spirit. The internet provides the foundation for formation of global communities around specific projects, and thanks to the proliferation of high-bandwidth connections, distribution costs have been reduced to near zero.

The PyMOL molecular graphics system is a concrete and successful example of this new paradigm for scientific software development. Here I explore how PyMOL will specifically help to spread the open-source vision, and how we will be able to assemble a complete platform for biomedical science through industry-supported development of critical components. Also addressed are some of the economic issues surrounding open-source development and an important role to be played by independent open-source software publishers.

About the speaker:

By day, Dr. Warren L. DeLano is an Informatics Scientist at Sunesis Pharmaceuticals, Inc. By night, he is a leading developer of open-source scientific software. His top project, PyMOL, is the most capable molecular graphics system available today as unrestricted open-source code. It is also one of the very few high-performance packages to support a native Python interface, which places unprecedented flexibility into the hands of the average research scientist.

DeLano's mission is to develop high-impact biopharmaceutical tools that directly hasten scientific and medical breakthroughs. At Yale, he worked for several years with Dr. Axel T. Brunger on the popular X-PLOR and CNS programs, which have been used to determine about half of all known biological structures. DeLano then went on to obtain a Biophysics Ph.D. in 1999 from the University of California, San Francisco. Through an unusual academic/industrial alliance, he performed the bulk of his graduate research at Genentech, Inc., working with Dr. James A. Wells, a pioneer of protein engineering. There, DeLano applied molecular biology, X-ray crystallography, and computational analysis to study the origins of binding energy in protein interfaces. He developed the RigiMOL program to visualize conformational changes, and SiteFinder to identify potential binding sites on protein surfaces.

In 1998, DeLano left Genentech to play an key role in the launch of Sunesis Pharmaceuticals, a company focused on exploiting energetic protein "hot spots" for purposes of drug discovery. For four years, he has led the development of informatics infrastructure at Sunesis, where open-source components have played a major role, from deployment of Linux servers and a Linux computing cluster, to creation of a Python-based chemical informatics platform. DeLano believes that the biotechnology and pharmaceutical industries have a significant and unrealized interest in supporting open-source software development, and he is now forming a consortium to fund open-source development of essential research tools.

Contact information:

Warren L. DeLano