Function Prediction

From High Throughput to Individual Proteins

 

This year, CASP, "the Olympic games of prediction,” which in the past focused only on predicting protein structure, will be a little different. The participants will be asked to predict not only the structure of each target sequence, but also its function. How could this be done? In this tutorial we will discuss the different approaches, tools and challenges in the emerging field of function prediction.

 

Most function prediction methods are designed for automated, high throughput prediction. But as in the case of CASP, a typical user is usually interested in a custom tailored prediction for a particular protein. We will present ways in which these high throughput methods could be used to predict the function of individual proteins. We will also survey the algorithms and scientific insights that underlie them.

 

The tutorial will elaborate on specific examples (including some targets from CASP 2004), and will discuss research opportunities and open challenges in the field.

 

Instructors: Our research is dedicated to in-silico study of protein structure and function. We have published papers and book chapters on different aspects of function prediction, and have taught related issues to audiences with variety of backgrounds.

 

Yanay Ofran, PhD - Researcher at Columbia University Bioinformatics Center (C2B2). Current research focuses mainly on analysis and prediction of protein-protein and protein-DNA interactions. Recently, coauthored a review about function prediction (2003, Automatic prediction of protein function, CMLS: 60(12):2637-50), and a book chapter about prediction methods using protein sequence (2004, Predictive methods using protein sequence, Bioinformatics, Baxevanis and Ouellette, Wiley Interscience). Teaching experience includes developing and teaching courses at graduate and undergraduate levels, as well as presentations and tutorials at scientific meetings and seminars.

 

Marco Punta, PhD - Researcher at the Columbia University Bioinformatics Center (C2B2). Current research focuses on predicting internal inter-residue contacts. PhD in Biophysics from the International School of Advanced Studies (ISAS) in Trieste, Italy; topic: comparative modeling and molecular dynamics simulations applied to the study of ion channels. Presentations at several meetings and seminars.

 

Burkard Rost, PhD - Positions: 1990-1994 EMBL Heidelberg, 1995 EBI Cambridge, 1996-1998 EMBL Heidelberg, 1998 LION Biosciences, 1999-now Associate Professor at Columbia University. Research focuses on methods predicting protein structure and function from sequence. The major goals are to develop tools that can be applied in the context of analyzing entirely sequenced organisms. Achievements in numbers: 92 invited talks in 16 countries; 101 papers published and submitted; first and last-author papers quoted over 5500 times; average citation of first and last-author papers over 110; online version of first Web server for prediction in molecular biology.