Expanding the limits of biomolecular simulations: revealing the mechanisms of blood clot formation using Fluctuating Finite Element Analysis.

Lead Research Organisation: University of Leeds
Department Name: Physics and Astronomy

Abstract

Computational simulation of biomolecules has proven to be a very useful approach during the past few decades, and is now considered essential in broad range of disciplines ranging from the molecular understanding of life to drug discovery. Molecular dynamics is so frequently used to calculate the dynamic behaviour of proteins at the atomistic level firstly due to the large number of protein crystal structures that are publicly available in the Protein Data Base, but also because this methodology is well established, and excellent software packages are freely available to the academic biomolecular sciences community. However, we are still far from simulating cellular dimensions and time scales of entire biological processes. This will not be solved by hardware improvements in the foreseeable future, especially as the continuous increase
in computational power is slowing and may come to an end. As a consequence, new methodologies are needed to reach longer time- and length-scales.

This fellowship proposes to join two cutting edge methodologies in coarse-grained protein modelling to overcome this situation. Specifically,
I will work with the Fluctuating Finite Element Analysis that models proteins as a non-rigid continuum subjected to thermal fluctuations, and the Multi-Scale Coarse-Graining method that aims to describe simplified molecular interactions using a physically-based bottom-up approach.

Once this methodology is ready, I will implement it within a scalable piece of software suitable for High Performance Computing, and will use this new tool to simulate the fibrin network self-assembly process, one of the key events in clot formation. This is a highly important biological system, as in vivo imbalance is related to a number of human pathologies, including heart and brain infarction. Structural data on the clot architecture has been shown to correlate with clinical data on cardiovascular diseases.

I will use currently available experimental data to demonstrate the capabilities of the proposed methodology and software. Next, further simulations will shed light on association pathways and affinities leading to fibrin polymerisation, on the process of lateral aggregation of protofibrils, on the role of each of the known interaction sites, on the influence of the external flow, and on the effect that some pathological mutants have on the self-assembly process and the final structure of the clot.

Full accomplishment of these objectives will result in significant advances in biomolecular modelling methodology, together with the release of a general purpose application for biomolecular simulations on the mesoscale, and medically relevant results on the clot formation process and structure.

Planned Impact

The proposed research will produce three main deliverables. Firstly, there will be simulation results revealing the mechanisms of the molecular basis of fibrin clot formation. Secondly, a scalable piece of software to simulate proteins at the mesoscale will be released. And thirdly, physics based methodology on protein-protein interaction modelling. The impact of these deliverables will target the Healthcare sector, essentially through pharmaceutical industry. The implicit generality of methodological research will enable me in the future to consider applications such as the discovery of new bio-materials and the design of vehicles for drug delivery. Both of these areas of biotechnology are active areas of experimental research at Leeds.

Understanding fibrin clot formation has direct clinical implications in bleeding and thrombotic complications of several diseases. For instance, commonly used medications to prevent thrombosis have turned out to have secondary effects on the final structure of the clot, resulting in a susceptible to lysis clot. On the contrary, there is a new generation of specific drugs that target and modulate the so called "knob-hole" interactions (see fig. 2 in the case for support), mostly avoiding such secondary effects. While the importance of these interaction sites have been revisited, the proposed research will be able to assess the consequences that the modulation of all the interaction sites involved in the fibrin self-assembly process, have on the overall mesh structure.

International pharmaceutical companies that can be potentially interested in the fibrin results include Enzyme Research Laboratories, GEHT, George King Bio-Medical Inc., Haematologic Technologies Inc., and Biopur while others like BAYER, novo nordisk, Grifols, Kedrion Biopharma, Octa Pharma, Immucor are known to have R&D departments more generally focused on Thrombosis and Haemostasis. More generally, in silico methods like Virtual Screening have been crucial to decrease the costs of the pharmaceutical research. The software and methodology proposed to parameterise the interactions will extensible to other protein systems, with applicability in the pre-discovery phase to obtain new compounds. In addition, it could be used to expand the amount of known targets, as currently 50% of them belong to 4 families: GPCRs, nuclear receptors, ligand-gated ion channels, voltage-gated ion channels. In a next future, it could be applied to model and design drug delivery liposomes, both with specific surface ligands, as well as with the desired rhelogical properties. In summary, the deliverables of the proposed fellowship will become directly relevant to the drug research and hence of interest for pharmaceutical industry.

As published in Forbes[1], the pharmaceutical industry destroyed nearly 300,000 jobs in the 2000 to 2011 period, and their efficiency in R&D has been declining during decades, with the number of new drugs per billion US dollar being reduced by a factor of 80 since 1950[2]. With an ageing society, increasingly dependent on new medicines, this has become a public major concern as is reflected in the EPSRC Healthcare
Technologies theme. The proposed research is an excellent exponent of the impact that basic research has on the health and quality of life.

References:
[1] M. Herper, A decade in drug industry layoffs, 2011, available at:
http://www.forbes.com/sites/matthewherper/2011/04/13/a-decade-in-drug-industry-layoffs.
[2] J. W. Scannell et al., Nat Rev Drug Discov 11, 191 (2012).

Publications


10 25 50
 
Description A first attempt to parallelise FFEA using MPI 
Organisation University of Edinburgh
Department Edinburgh Parallel Computing Centre (EPCC)
Country United Kingdom of Great Britain & Northern Ireland (UK) 
Sector Academic/University 
PI Contribution I helped the student to compile FFEA on Archer, wrote FFEA documentation on usage, development and installation, and prepared a new system (a number of input files) to be used to find performance bottlenecks in FFEA and to benchmark a future parallel implementation of the FFEA.
Collaborator Contribution Guanhao Lu is a new master student at EPCC under the supervision of Toni Collis. He has been familiarising with FFEA, and he is starting working on is project imminently.
Impact A first MPI version of FFEA was delivered, its performance discussed in Guanhao Lu's MSc dissertation: https://static.ph.ed.ac.uk/dissertations/hpc-msc/2015-2016/Guanhao_Lu-MSc-dissertation.pdf The resulting version of the code is able to run on multiple nodes (different computers) in parallel, although it does not scale well, because only one section was parallelised. Nonetheless, this first approach served as the basis for a new proposal to parallelise FFEA. More explicitly, Sarah Harris, Toni Collis, and myself put in an eCSE proposal: http://www.archer.ac.uk/community/eCSE/ that was submitted recently with letters of support from CCP-BioSim and CCP-EM communities.
Start Year 2016
 
Description FFEA performance assessment 
Organisation Numerical Algorithms Group Ltd
Country United Kingdom of Great Britain & Northern Ireland (UK) 
Sector Private 
PI Contribution I submitted a service request to Performance Optimisation and Productivity (POP): https://pop-coe.eu/request-service-form to audit the performance of FFEA. I prepared a benchmark (a set of input files to simulate a molecular system) ensuring that it was stressing all parts of the code, so that the performance analysis would report on every different function. I was also introducing the usage of FFEA, and helping with some related problems. I addressed most of the issues raised by the resulting report from POP, increasing the performance of our code.
Collaborator Contribution Jonathan Boyle in collaboration with Nick Dingle and Sally Bridgwater, all of them working at NAG, run a number of simulations using the setup I provided studying the scalability, efficiency, computational performance and load balance.
Impact A report on the computational performance of FFEA was delivered by the team at NAG, pointing to some bottlenecks. Consistently, I tackled these, writing a faster and more robust version of FFEA that will be released very soon under a GPL licence. The collaboration has not ceased, and a new performance analysis could be done by the team at NAG providing guidance to a further step in the performance of FFEA.
Start Year 2016
 
Description Investigating the parallel performance of the Fluctuating Finite Element Analysis tool 
Organisation University of Edinburgh
Department Edinburgh Parallel Computing Centre (EPCC)
Country United Kingdom of Great Britain & Northern Ireland (UK) 
Sector Academic/University 
PI Contribution I provided many explanations, and a number of input files to simulate different systems with different characteristics, making easy to Jana Boltersdorf to benchmark and study the computational performance of the FFEA code. Weekly meetings were held to help.
Collaborator Contribution Jana Boltersdorf, who was a summer student at the EPCC, worked on characterising the performance of the FFEA code, which is being developed in our group, and provided a series of speed up plots as well as a final report.
Impact - A series of speed-up charts where the run time of FFEA was measured using an increasing number of processors
Start Year 2015
 
Description Visualisation of proteins using Paraview 
Organisation University of Edinburgh
Department Edinburgh Parallel Computing Centre (EPCC)
Country United Kingdom of Great Britain & Northern Ireland (UK) 
Sector Academic/University 
PI Contribution I provided a number of challenges on visualising FFEA trajectories using Paraview. Using Paraview (http://www.paraview.org/) was proposed after I found our current viewer to perform poorly on medium and large sized systems, and after finding the limitations that VMD (http://www.ks.uiuc.edu/Research/vmd/) had in handling our sort of trajectories. In addition, I provided a number of systems to be visualised (input files) and weekly meetings were held sharing ideas and giving advice.
Collaborator Contribution Ondrej Vysocký, who was a summer student at EPCC under the supervision of Dr Toni Collis and Dr Neelofer Banglawala , wrote a python script to convert FFEA trajectories into VTK format. They could then be read by Paraview, and Ondrej provided with a number of recipes to display different attributes, making easier to visualise simulation output trajectories.
Impact - A FFEA to VTK file converter written in Python. - A number of recipes to visualise VTK trajectories using Paraview.
Start Year 2015
 
Description Advanced Research Computing User Meeting (Leeds) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Professional Practitioners
Results and Impact I gave a talk entitled "Simple tricks to improve the performance: a real case study" in the Advanced Research Computing User Group. The meeting brings periodically together the users of the main computer cluster of the University of Leeds, and in that talk I showed what I had done, with minimal effort to improve the performance of the code that I use to simulate proteins dynamics. I was the first user reporting and showing the usage of the Allinea MAP, a powerful software profiler that the University had purchased recently, motivating new users to try it. In addition, there was a nice discussion on other approaches I could considerate to improve even more the performance of the code at a low cost.
Year(s) Of Engagement Activity 2015
 
Description Astbury Conversation (Leeds) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact I presented a poster at this short conference engaging 250 scientist from all over the world. People who I talk to includes Nobel laureate Michael Levitt who made reference to our work later in his talk.
Year(s) Of Engagement Activity 2016
URL http://www.astburyconversation.leeds.ac.uk/symposium.php
 
Description CCPBioSim/CCP5 Multiscale Modelling Conference (Manchester) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact The conference brought together international leaders in a number of research areas, from fundamental physics to materials and biology, sharing interest in multi-scale modelling. I presented a poster in the conference, which was well received, and the discussion arising from the very creative and different approach was truly enriching.
Year(s) Of Engagement Activity 2016
URL https://eventbooking.stfc.ac.uk/news-events/second-ccpbiosimccp5-conference
 
Description Combining Simulation and Experiment in Molecular Imaging (Southampton) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact I presented FFEA, the modelling approach that we are developing in Leeds, to an audience mainly composed of postgraduate students with a biology background. The main aim was to show how the latest low resolution imaging techniques can be combined with molecular simulation.
Year(s) Of Engagement Activity 2016
URL http://www.southampton.ac.uk/ifls/news/events/2016/05/25-molecular-imaging.page
 
Description European Biophysical Society Association Congress (Dresden) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact I presented a poster entitled "Understanding fibrin protofibrils at the molecular level through a new coarse-grained approach" at the European Biophysical Society Association Congress. There were several very interesting discussions, some of them methodological and explicitly on different coarse-grained approaches (so in the field that I am doing research) and some of them on fundamental biophysical processes that would benefit enormously if using the meso-scale approach that I am developing. Extremely useful.
Year(s) Of Engagement Activity 2015
URL http://ebsa.org/ebsa-2015
 
Description International Fibrinogen Workshop (Skukuza, South Africa) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact The 4 days long fibrinogen workshop brought together the leading researchers of the field in an ideal environment for discussion and collaboration. I was presenting a poster on the types of calculations and modelling that we are working on, and it certainly was well received. A number of ideas came to me, and conversations for establishing future collaborations were started.
Year(s) Of Engagement Activity 2016
URL http://www.nwu.ac.za/fibrinogen
 
Description International on Thrombosis and Haemostasis (Toronto) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact I presented a poster entitled "Understanding Fibrin Protofibrils at the Molecular Level: Formation and Flexibility Through a New Coarse-Grained Protein Modelling Approach". The poster had a great reception as the conference attendees were mostly experimentalists, and most of them were working on larger and much larger time and length-scales. It was very stimulating to meet the experts that could use the theoretical approaches that I am developing, and see that they had high interest in my work.

The same trip to Toronto was used to meet Régis Pòmes at the Sick Kids Hospital:
http://biochemistry.utoronto.ca/person/regis-pomes/
to talk about future collaboration using FFEA to model a number of cellular processes involving aggregation.
Year(s) Of Engagement Activity 2015
URL http://www.isth.org/mpage/2015Microsite
 
Description N8-HPC Multiscale Computational Mechanics (Sheffield) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Professional Practitioners
Results and Impact I was selected to give an oral presentation entitled "A bottom-up approach to describe protein-protein interactions in a continuum mechanics biomolecular model" in this "Network Event" with the aim of bringing the regional researchers in Multiscale Computational Mechanics together. It was a very useful meeting as most of the attendees were engineers, and that gave me the opportunity to hear about a wide range of approaches used on different systems. In addition, the event was organised by the N8-HPC, which brought in the computational perspective. Truly multidisciplinary.
Year(s) Of Engagement Activity 2015
URL http://n8hpc.org.uk/n8-hpc-network-event-multiscale-computational-mechanics/
 
Description Workshop in Molecular Modelling and Simulation (Leeds) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact I gave a talk presenting our multi-scale model for molecular biology in the department of Chemistry of the University of Leeds, where Prof Carol Hall from North Carolina State University also presented her works on coarse-grained molecular modelling. The audience background was mixed, mostly from biology or chemistry. There was a very interesting final discussion involving Andrew Wilson from Chemistry, Carol Hall, Sarah Harris (from physics) and myself.
Year(s) Of Engagement Activity 2016
 
Description Workshop on Biophysical Modelling and Simulation (UCL, London) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact This was a small single-day workshop where people from Peter Coveney's (UCL) and from Sarah Harris' (Leeds) groups presented their current work in a number of short talks. I gave a talk on the methodological work that I am doing, raising a great interest. There was a very interesting discussion about the convergence of the methodology with a direct effect on the decisions that I was taking afterwards on my research. In addition, Peter Coveney introduced his works on fluctuating hydrodynamics, which could help in the future at introducing hydrodynamics in our FFEA model, leading to future collaboration.
Year(s) Of Engagement Activity 2015