Centre for Computational Science: Biological and Biomedicine research highlights
Director: Professor Peter Coveney
• Centre established 2003
• Involved in many UK, EU and international projects.
• Approximately 20 full time members
• Range of problems: theoretical & computational science,
computer science, distributed computing, visualisation and
advanced networking.
• Our different computational techniques span time and lengthscales from the macro-, through the meso- and to the nano- and
microscales.
• We are committed to studying new approaches and
techniques that bridge these scales.
• For more information see:
http://ccs.chem.ucl.ac.uk
Modelling for Clinical Decision Support: HIV Protease
Simulation for Surgeons: Cerebral Blood Flow
Our
tool,
HemeLB,
simulates the flow in a
region of interest and
interactively
generates
images of flow velocity,
pressure and vessel wall
stress (such as shown to
the left). These may
eventually help surgeons
make more informed
decisions as to if and
how to intervene.
1
Personal genetic
sequence data soon
available as we head
towards $100
Genome Sequencing
3
cctcaaatcactctttggcaacgaccc
atcgtcacaataaagataggagggc
aactaagggaagctctattagataca
ggagcagatgatacagtattagaaga
cataaatttaccaggaagatggaaac
caaaaatgatagggggaattggaggt
tttgccaaagtaagacagtatgatcag
atacccgtagaaatctgtggacataaa
gttataggtacagtattagtagg
ranking of drug
binding affinities
The causes of neurovascular pathologies (such as intracranial aneurysms)
are not well understood but are likely to include properties of neurovascular
blood flow.
Simulation
Wild-type & mutants
LPV
2
A schematic representation of the process is shown above: vessel
geometries are reconstructed from medical imaging data (typically
CT rotational angiography), which are then input to the fluid flow
simulation and results returned to inform decision making.
Catalytic ASP
dyad
Build patient
specific models
of receptors
Rank binding
affinity against
available drugs
Select the one
best suited to
the individual
4
HPC allows atomic
simulation on clinically
relevant timescales
Include in clinical decision
support system
Mutations
In order to be of diagnostic relevance, the tool we aim to create
must be fast and accurate. Our current work is focused on
demonstrably improving these aspects. To the right are two
images of the velocity field error (compared to high-resolution
CFX simulation) in a realistic vessel geometry, before (left) and
after (right) preliminary work.
Sequence interpretation already routine in HIV treatment. Ideal examplar system to develop clinically relevant models. The
same methodology can be applied to the treatment of other infectious diseases and medical conditions such as cancer.
I. Stoica, S. K. Sadiq, and P. V. Coveney, JACS, 130, (8), 2639-2648, (2008)
Drug Efficacy and Kinase Activation: Epidermal Growth Factor Receptor (EGFR)
EGFR is a major target for drugs in treating lung carcinoma since it promotes cell growth and tumour progression. Its mutations promote
conformational changes which affect its activation and drug efficacy. we have studied changes in drug binding affinities due to cancer
mutations of EGFR using ensemble molecular dynamics simulations, and addressed
an enhanced activation mechanism thought to be associated with mutations.
Active
Inactive
state
state
Translocation of Polynucleotides through Protein Nanopores
Polynucleotide (e.g. single stranded DNA) translocation
through confined protein-pores is a system with
experimental relevance in single channel current
recording (SCCR, see right), which is a candidate for next
gen genome sequencing.
A
We can use simulations to gain insight, providing knowledge to improve SCCR to the
point of genetic sequencing. Since the system is complex and represents relatively
large time scales, we speed up the process in order to allow highly detailed, atomistic,
simulations.
Binding Free
Energy
Calculator
Binding Affinity Calculator (BAC) is used to determine patient
specific response to lung cancer drugs. The workflow is fully
automated for execution of each component and transfer of files
across distributed HPC resources.
Methods of rapid translocation include cv-SMD (left) or ABF (right).
Mapping the conformational free energies of wild-type and
mutant EGFRs onto a principal component space enables
identification of population changes in various conformations
upon mutation,
EGFR mutation introduces conformational changes in the active and inactive states. These mutation-associated changes alter the drug
efficacy by distorting the ATP- and drug-binding cleft, and adjust the activation of the EGFR kinase by affecting the relative stabilities of the
active and inactive conformations.
S. Wan and P. V. Coveney, J R Soc interface (2011) DOI: 10.1098/rsif.2010.0609
(right)
Our
simulations
reproduce the experimental
finding that polyadenine (“a”)
translocation more slowly
than polydeoxycytidine (“b”)
(right) We confirm the observed trend,
and provide additional insight: for
example, a phosphate-lysine interaction
between the DNA and protein, and a
contribution of base-stacking