Computational biology and computational biologists Tandy Warnow UTAustin

Computational biology and computational biologists Tandy Warnow, UT-Austin Department of Computer Sciences Institute for Cellular and Molecular Biology Program in Evolution, Ecology, and Behavior Center for Computational Biology and Bioinformatics

Two computational biologists • One computational biologist needs to know a lot of biology • Another needs to know a lot of mathematics

Another two computational biologists • Craig Benham: mathematics of stressed DNA (understanding regulation) • Gene Myers: whole genome sequencing and BLAST

Two different types of computational biologists • One works on mathematical or computational problems (derived from biology) that are well posed, and are hard to solve -- these need significant computer science/math/statistics • One works on biological problems that are not well posed, and where the computer science/math/statistics needed may be “easier” • Both can be problems that are important to biologists, and which they cannot solve without computational biologists’ involvement

My view of Pasteur’s Quadrant Hard math Easy math Easily applicable Not applicable

My view of Pasteur’s Quadrant Hard math Easy math What computational scientists want Easily applicable Not applicable

My view of Pasteur’s Quadrant Hard math Easy math What computational scientists want Easily applicable What computational scientists do Not applicable

My view of Pasteur’s Quadrant Hard math What computational scientists want What computational scientists do What biologists want Easy math Easily applicable Not applicable

Phylogeny From the Tree of the Life Website, University of Arizona Orangutan Gorilla Chimpanzee Human

DNA Sequence Evolution -3 mil yrs AAGACTT AAGGCCT AGGGCAT TAGCCCA -2 mil yrs TGGACTT TAGACTT AGCACAA AGCGCTT -1 mil yrs today

Molecular Systematics U AGGGCAT V W TAGCCCA X TAGACTT Y TGCACAA X U Y V W TGCGCTT

Computational challenges for Assembling the Tree of Life 8 million species for the Tree of Life -- cannot currently analyze more than a few hundred (and even this can take years) • We need new methods for inferring large phylogenies - hard optimization problems! • We need new software for visualizing large trees • We need new database technology • Not all phylogenies are trees, so we need methods for inferring phylogenetic networks

Time is a bottleneck for MP and ML • Systematists tend to prefer trees with the optimal maximum parsimony score or optimal maximum likelihood score; however, both problems are hard to solve • (Our experimental studies show that polynomial time methods do not do as well as MP or ML heuristics, when trees are big and have high rates of evolution) Local optimum MP score Global optimum Phylogenetic trees

MP/ML heuristics Fake study Performance of hill-climbing heuristic MP score of best trees Time

DCM-boosting Speeding up MP/ML heuristics Fake study Performance of hill-climbing heuristic MP score of best trees Desired Performance Time

Characteristics • The research can be published in mathematics/statistics/computer science journals and conferences, and evaluated along these lines • These people can be faculty in Math/Statistics/Computer Science departments, and *maybe* in some biology departments • Substantive improvements are hard, but if achieved will have enormous impact on many biologists • Why? These are old problems, endorsed by biologists, of a computational nature.

The “other” type • Deals with problems like: protein fold prediction, inferring metabolic or regulatory networks, finding genes within genomes, or even computing a good multiple sequence alignment • Needs to know a lot of biology to pose appropriate computational problems • Resultant algorithms may not (in some cases) make for interesting or publishable mathematics • Note: generally new problems because of new data

What’s needed (for all types) • Ability to collaborate with a variety of people, and learn what they want to achieve • Ability to be flexible in terms of how one evaluates research results (e. g. , real vs. simulated data, theory versus experiment) • Ability to communicate research results to different types of researchers • Ability to use a variety of techniques to solve biological problems • Ability to model and pose appropriate computational approaches for biological problems

Difficult questions • What departments should have computational biologists (especially of the second type)? • Should there be departments of computational biology? • Should there be Ph. D programs in computational biology? • How to evaluate a computational biologist of either type?

Some issues for academic computational biologists • Journal versus conference papers, and number of each • Experimental/empirical versus theoretical work • Software versus papers • Authorship order within publications • Promotion and Tenure in two departments? • Biggest issue: How to predict future success? ? ?