Detecting channels
in proteins
Ondřej Strnad
outline
• biochemical background
• what a channel in a protein is
• what we have so far
• what another directions should be (in relation to my PhD.
thesis)
• how are our algorithms available to common public
• conclusion
14.3.2011
2
biochemical background
• protein molecule
• for us thousands of atoms (spheres) with different radii (diferent
chemical item)
• substrate
• small protein molecule (tens of atoms)
14.3.2011
3
channel
• union of spheres with centres on the path that connect an
active site with the boundary of the molecule and not
intersect any atom
• channels are used by substrate to penetrate into protein
• existence of channels provide to chemists important
information about protein behaviour and may emphasise
places where the empty space is
• substrate penetration is important for instance during drug
design
14.3.2011
4
channel detection algorithms
• based on computational geometry
• utilizing Voronoi diagram and it’s dual Delaunay triangulation
• and Dijkstra’s algorithm (maximizing cost function)
14.3.2011
5
channels in a single snapshot
• we are able to detect channels in one snapshot (stored
positions of all atoms in given time)
• a number of detecting channels is defined by the user
14.3.2011
6
biochemical background
• in real life atoms in the protein do not remain still
14.3.2011
7
channels in molecular dynamics
14.3.2011
8
• sequence of snapshots – sampling frequency in [ns]
clustering
• more channels may seem to be similar – there may be only
small differences in their path in different snapshots
• two channels which are closer than user-defined threshold are
marked as similar and as members of one cluster
• clustering gives additional information about the parts in the
protein where empty space is pulsing
14.3.2011
9
channels in molecular dynamics
14.3.2011
10
channels in molecular dynamics
1) create initial graph from a given snapshot
2) process another snapshot
• track all edges
• update the edge value where necessary
3) process Dijkstra’s algorithm
• number of edges remain the same
• emphasize pulsing parts
BENEŠ Petr, MEDEK Petr, STRNAD Ondřej, SOCHOR Jiří - Computation of
Dynamic Channels in Proteins [BIOTECHNO 2011]
14.3.2011
11
channels in molecular dynamics –
constrained channels
• building one large multi-edge graph
• generate huge amount of edges
• defining constraints in order to decrease the total number of
traversing edges
BENEŠ Petr, STRNAD Ondřej, SOCHOR Jiří, New path planning method for
computation of constrained dynamic channels in proteins [WSCG 2011]
14.3.2011
12
current research
• continue with validation of biochemical relevance on different
proteins
• with the increase of computational power, the chemists are
capturing larger and larger proteins and longer sequence of
snapshots
• try to investigate behaviour of our algorithms on very large
proteins (hundreds of thousands of atoms) - optimization
14.3.2011
13
another identified directions of research
• include some biochemical properties during the channel
computation (van der Waals forces, …)
• channels with different shape of cross-section – maximizing
the channel
• put the simulation of penetration of substrate into haptic
environment (currently solved by Dr. Křenek and his team) and
use it with our algorithms
14.3.2011
14
availability of our algorithms
• Caver viewer
• software developed by Caver team (HCI and Loschmidt
laboratories)
• available as Java-applet (www.caver.cz)
• Caver plugin for PyMOL
14.3.2011
15
conclusion
• detecting channels in proteins is very important for
biochemists
• we are able to detect channels in the single snapshot or
sequence of snapshosts
• nowadays we are focusing on molecules with large amount of
atoms
• results of our work are accessible for general public via Caver
Viewer software
14.3.2011
16
conclusion
• detecting channels in proteins is very important for
biochemists
• we are able to detect channels in the single snapshot or
sequence of snapshosts
• nowadays we are focusing on molecules with large amount of
atoms
• results of our work are accessible for general public via Caver
Viewer software
Questions?
14.3.2011
17