Folding@home and SWISS-MODEL

Taru Tukiainen ja Sini Sipponen

S-114.250013.12.2006

Two Different Approaches to Protein Structure Modeling

Outline Introduction to protein modelling Modelling the folding process with

Folding@home Comparative modelling with SWISS-MODEL

Introduction

Proteins are formed from a sequence of amino acids

Primary structure = polypeptide chain

Secondary structure alpha helix beta sheet

Tertiary structure is 3D Quaternary structure is comprised

of several tertiary structures Native state is the functional form

Hydrophobic effect on Folding

Important force affecting the forming of the tertiary structure

Many residues of amino acids are hydrophobic

leads to the formation of a hydrophobic core

After folding, entropy of protein decreases, but the entropy of system decreases

Entropy promotes the folding process

Misfolding

Prions are result of faulty folding

Little known about how they form

Can convert normal protein molecules to prions

The Problems

1. Impossible to predict 3D structures from polypeptide chains

2. Folding processes and mechanisms are mostly unknown

3. the 3D, native state is very expensive and time consuming to solve

Modeling the Folding Process

Proteins fold in about 10 µs Simulation would take dozens of years

Proteins are formed of thousands of atoms Presented often as force fields

E.g. temperature, pH, covalent bonds between residues and hydrophobic effect must be taken in consideration

Proposed that the native state = minimum of

potential energy curve

Folding@home Objective to study the

dynamics of protein folding and misfolding and the ensuing diseases

Uses distributed computing, volunteers let their PC to be utilized when they aren’t needed

Program is a screen saver Uses packages

AMBER, TINKER and GROMAS

Ensemble Dynamics Method

Polypeptide chain is considered as a system waiting for enough free energy to overcome the free energy barrier (= the folding)

Group of several molecules M is simulated at the same time simulation rate is then M times faster than a single

simulation The simulations are completed in hours not in years

Wait until the first one of the simulations overcomes the energy barrier All the simulations are restarted from the new energy level

Comparative modelling

aims at building a 3D model for a protein with unknown structure

relies on detectable similarities between the protein sequence being modelled (the target) and at least one empirically determined protein structure (the template)

a small change in the protein sequence usually results only in a small change in its 3D structure

SWISS-MODEL fully automated web-server for protein

structure modelling developed in 1993 nowadays the most widely-used free web-

based automated facility

Using SWISS-MODEL User-friendly User only submits the amino acid sequnce

on a web form optionally templates can be submitted as well

Results in 15-60 min by e-mail

1Search for suitable templates

2Check sequence identity with target

3Create ProModII jobs

4Generate models with ProModII

5Energy minimisation with Gromos96

 

First Approach Mode (regular)  

First Approach Mode (with user-defined templates)  

Optimise Mode  

How SWISS-MODEL works? Five steps that can be repated iteratively

How SWISS-MODEL works? Step 1

search for suitable templates from ExNRL3D program used: BLASTP2

Step 2 find sequences with good degree of similarity

(>25%) aling target and template sequences program used: SIM

How SWISS-MODEL works? Step 3

create ProModII input files Step 4

generate models program used: ProModII

Step 5 minimize energy program used: Gromos96

Are there problems with SWISS-MODEL? Results must be concidered with care

procedure is non-experimental no human intervention during model building

Chosen template affects the results the more the template and the target

sequence share identity the more accurate the results will be

Accuracy of SWISS-MODEL