eScience Seminar: Why Computer Scientists Should and Can Learn Biochemistry

Date: 
May 22, 2013 | 4:00 pm - 5:00 pm
Location: 
EE 303

Joseph Hellerstein is Manager of Computational Discovery for Science at Google. In Autumn Quarter 2013, he will be teaching a seminar at UW entitled "Biochemistry for Computer Scientists." The seminar will provide a view of biochemistry as a mix of organic and information science. Joseph's talk on May 22 will provide an in-depth introduction to this upcoming seminar.

Abstract

The advent of inexpensive DNA sequencing technology has created wide ranging opportunities in the science and engineering of life processes. Training for these opportunities requires a blend of computer science, statistics, and biochemistry. Some effort has been devoted to training biochemists in software engineering. This talk advocates educating computer scientists in biochemistry and provides a glimpse into the course specifics.

Typically, the prerequisites for introductory biochemistry are general chemistry (1 year), organic chemistry (1 to 2 semesters), and other related sciences. This is formidable background for a computer scientist with little training in physical sciences.

The thesis of this talk is that much of biochemistry is about information science. I claim that only a modest background in organic chemistry is required to understand the basics of biochemistry, which can be covered in a single lecture. For example, the background includes numbering Carbons of organic molecules and identifying a few chemical functional groups.

A bigger challenge in the biochemistry of life processes relates to information science, not chemistry. To elaborate, understanding life processes requires insights into molecular structures (and hence function) and reaction pathways. Molecular structures can be described using CS techniques such as class diagrams. And, reaction pathways can be expressed as workflows that transform molecular structures by the use of enzymes. However, applying CS techniques to biochemistry does require some adaptation. For example, computer science workflows have unidirectional steps from inputs to outputs. In chemistry, reactions are bidirectional, and so the workflow representation is bidirectional. A second adaptation of CS tools relates to regulation of reaction pathways by changes to enzymes (e.g., allosteric regulation and covalent modification). Regulation is easily represented by workflows that act on enzymes. However, unlike CS workflows, in chemistry the behavior of a step may change if other parts of the workflow modify the enzyme used in that step.

In sum, I claim that students with a strong CS background and only a cursory knowledge of general chemistry can learn core elements of biochemistry in a 10 week quarter.

Joseph Hellerstein CV | LinkedIn

AttachmentSize
UW eScience - Why Computer Scientists Should and Can Learn Biochemistry.pdf736.86 KB