Title:	Experimentally Guided Modeling of Protein Complexes: Application to disease-related proteins
DNr:	SNIC 2015/1-121
Project Type:	SNIC Medium Compute
Principal Investigator:	Björn Wallner <bjorn.wallner@liu.se>
Affiliation:	Linköpings universitet
Duration:	2015-05-01 – 2016-05-01
Classification:	10601 10203 10610
Keywords:

Abstract

Because triolith was super crowded last year I was only granted 80k hours/month, which was way too small for the ambitious goals in the granted VR project. Hopefully the situation have improved now with the installation of the new PDC system. I sent in an application two month ago and was granted an additional 55k (2015/1-83) until the end of this project, which is now up for renewal. Ideally, the projects could be merged into one larger allocation. Project: Proteins are key players in virtually all biological events that take place within and between cells. They often accomplish their function as part of larger protein complexes. Unfortunately, compared to structure determination of individual proteins, structural characterization of large macromolecular assemblies is much more difficult and no single experimental method can accomplish this task alone. In this proposal we suggest an integrative approach to combine all available experimental information about a given assembly using computational modeling. To this end we will develop an INTEGRATIVE MODELING FRAMEWORK capable of handling the different resolution ranges for the different types of experimental data, ranging from putative interaction data all the way to atomic X-ray structures. Even though this is ultimate goal, we will initially study interacting proteins in complex purely by computational means. The reasoning is that incorporating experimental data at the end can only make the result better and by focusing on the more difficult problem without experimental information it is clear that any improvements can be attributed the computational algorithm. As part of this we will also develop IMPROVED SCORING and SAMPLING TECHNIQUES for DOCKING. Finally, we will apply the developed methodology combined with existing tools to several disease-related proteins for which we have first hand access to experimental data, including TTR, TRIM21 and Myc. This work involves serious computational power; we will continue develop our own software, which will be a crucial part of the modeling framework. It will also involve large benchmarks to compare performance to existing tools. Just to give a ballpark figure, two benchmarks we are about to conduct in the coming month involve 200,000 core hours each.