Modelling Proteomes
Ram Samudrala
University of Washington
Overview
• De novo structure prediction: 3-6 Å RMSD up to 150 residues
A
• Comparative modelling: 1-4 Å RMSD for 50-20% identity
• Predict function based on structure, in conjunction with available
experimental data
• Predict structures and functions for all tractable proteins
• Integrate resulting data with data from single molecule as well
as genomic/proteomic studies: Bioverse
Bioverse – biology dataflow
Jason Mcdermott
Bioverse - information dataflow
Jason Mcdermott
Bioverse – explore relationships among molecules and systems
http://bioverse.compbio.washington.edu
Jason Mcdermott
Bioverse – explore relationships among molecules and systems
Jason Mcdermott
Bioverse – human protein-protein interaction network
Jason Mcdermott/Zach Frazier
Bioverse – salmonella protein-protein interaction network
Jason Mcdermott/Zach Frazier
Bioverse – human protein-protein similarity network
Jason Mcdermott/Zach Frazier
Bioverse – mapping pathways on networks
Inisitol phosphate metabolism
Benzoate degradation
Sphningoglycolipid metabolism
Starch/sucrose metabolism
Nicotinate and nicotinamide metabolism
Jason Mcdermott
Bioverse – mapping pathways on networks
Inisitol phosphate metabolism
Benzoate degradation
Sphningoglycolipid metabolism
Starch/sucrose metabolism
Nicotinate and nicotinamide metabolism
Jason Mcdermott
Bioverse – mapping pathways on networks
Oxidative phosphorylation
Jason Mcdermott
Bioverse – mapping pathways on networks
Oxidative phosphorylation
Jason Mcdermott
Bioverse – viewer
Galen Wilkerson
Bioverse – issues
• Algorithms – graph theory algorithms (connected components,
cliques, articulation points, condensation)
• Integration – how to normalise and weight different combinations of
data (neural networks, SOMs)
• Database – flexible searching of large amounts of data (OODBMS)
Take home message
Prediction of protein structure and function can
be used to model whole genomes to understand
organismal function and evolution
Acknowledgements
Group members