Modelling proteomes: Application to understanding HIV disease progression Ram Samudrala Department of Microbiology University of Washington How does the genome of an organism specify its behaviour and characteristics?

Proteome – all proteins of a particular system ~60,000 in human ~4500 in bacteria like Salmonella and E. coli Several thousand distinct sequence families 15 in HIV

Modelling proteomes – understand the structure of individual proteins A few thousand distinct structural folds

Modelling proteomes – understand their individual functions Thousands of possible functions

Modelling proteomes – understand their expression Different expression patterns based on time and location

Modelling proteomes – understand their interactions Interactions and expression patterns are interdependent with structure and function

CASP6 prediction (model1) for T Å C α RMSD for all 53 residues Ling-Hong Hung/Shing-Chung Ngan

CASP6 prediction (model1) for T Å C α RMSD for all 70 residues

Tianyun Liu CASP6 prediction (model1) for T Å C α RMSD for all 137 residues (80% ID)

CASP6 prediction (model1) for T Å C α RMSD for all 142 residues (46% ID) Tianyun Liu

oxidoreductase transferasehydrolaseligaselyase Similar global sequence or structure does not imply similar function TIM barrel proteins 2246 with known structure

Function prediction from structure Kai Wang

Prediction of protein interaction networks Jason McDermott Interacting protein database protein a protein b experimentally determined interaction Target proteome protein A 85% predicted interaction protein B 90% Assign confidence based on similarity and strength of interaction Key paradigm is the use of homology to transfer information across organisms; not limited to yeast, fly, and worm Consensus of interactions helps with confidence assignments

E. coli predicted protein interaction network Jason McDermott

H. sapiens predicted protein interaction network Jason McDermott

Identification of virulence factors

Bioverse – explore relationships among molecules and systems Jason McDermott/Michal Guerquin/Zach Frazier

Prediction of HIV-1 protease-inhibitor binding energies with MD Ekachai Jenwitheesuk Jenwitheesuk E, Samudrala R. Antiviral Therapy 10: , Jenwitheesuk E, Samudrala R. BMC Structural Biology 3: 2, 2003.

Prediction of HIV-1 protease IC 50 values with linear regression Kai Wang, John Mittler Prediction Experiment Wang K, Samudrala R, Mittler J. Journal of Infectious Diseases 190: , Wang K, Samudrala R, Mittler J. Antiviral Therapy 9: , Wang K, Jenwitheesuk E, Samudrala R, Mittler J. Antiviral Therapy 9: , Wang K, Samudrala R, Mittler J. Journal of Clinical Microbiology 42: , 2004.

Prediction of inhibitor resistance/susceptibility Ekachai Jenwitheesuk/ Kai Wang/John Mittler Jenwitheesuk E, Wang K, Mittler J, Samudrala R. AIDS 18: , Jenwitheesuk E, Wang K, Mittler J, Samudrala R. Trends in Microbiology 13: , 2005.

Prediction of fitness in absence of drug John Mittler Mittler J, Samudrala R. submitted, 2005.

Is amino acid change more than one nucleotide mutation away from query sequence? Exclude Yes Is this amino acid ever observed as a natural polymorphism in untreated patients? No Yes Is this amino acid in a database of mutants found to have nonzero fitness in in vitro studies such as those of Loeb et al.? Yes Is the new protein predicted to be nonfunctional based on all-atom score Yes No Include No Does amino acid change have a high Blosum62 or PAM score? Yes Include Exclude Prediction of escape mutants (HIV evolution) John Mittler

Acknowledgements Aaron Chang David Nickle Ekachai Jenwitheesuk Gong Cheng Jason McDermott Jeremy Horst Kai Wang Ling-Hong Hung Mike Inouye Michal Guerquin Stewart Moughon Shing-Chung Ngan Tianyun Liu Zach Frazier National Institutes of Health National Science Foundation Searle Scholars Program (Kinship Foundation) UW Advanced Technology Initiative in Infectious Diseases John Mittler and Jim Mullins