Computational Challenges in Biochemical Networks: Multiscale Modelling and Inverse Problems

Lead Research Organisation: University of Manchester
Department Name: Mathematics

Abstract

In the last 20 years, technologies have been developed which allow biologists to observe, in real time, the reactions which are occurring in a single cell. These new developments have the potential to give us a whole new understanding of how cells function. In particular, it gives us a tool with which we can make great leaps in our understanding of how our genes operate and affect the way cells behave, multiply, and die. The understanding of these mechanisms is key in developing treatments for conditions for when they go wrong, for instance in cancer. As such, this relatively young area of science could be very important for the future of the health of humankind.

The use of these technologies is increasing rapidly amongst biologists, but the problem remains as how best to interpret this data, and allow us to understand what we have observed. What is more, the observational methods are far from perfect, and are full of small errors which could cloud our conclusions. Therefore it is important that we understand the underlying mathematics within these problems, in a bid to extract as much reliable information from this data as possible.

The aim of this project is to study the mathematical theory behind, and develop new computer algorithms for, the analysis of this type of data. The Bayesian philosophy is a mathematical framework which allows us to not only identify likely biochemical mechanisms which could have caused the phenomena we observe in the experiments, but also to quantify how much we should believe our own results. This project will have significant impact on this area, and help to cement the UK's position as one of the leading places to conduct biological and pharmaceutical research, which plays such an important part in our economy. Furthermore, it will enhance the UK's reputation for high quality interdisciplinary applied mathematics research.

Planned Impact

This project is motivated by a genuine need within the biotechnology sector to be able to process and make the best use of the huge amounts of data that are emerging from new technologies. The technologies in question allow biologists to observe the biochemical reactions which are occurring in a single cell, in real time. The analysis of this data is a crucial step in furthering our understanding of how our genes alter the way cells behave, proliferate, and die. A greater understanding of how things can go wrong in cells is key to the development of new treatments for a whole range of different conditions. What is more, it will aid the development of the science and technology behind individualised treatments for conditions such as cancer, leading to far better survival rates.

Therefore, this project has a clear objective in impact, in terms of helping biologists and pharmaceutical researchers to further their understanding of biological reactions and then go on to develop new treatments. This will lead to several significant benefits to the UK in the long term. New treatments will help people to live longer, healthier and more active lives in future generations. The average life expectancy in developed countries has soared in the last 100 years, not least because of the advances in healthcare and pharmaceuticals. In order for this improvement in the quality and length of people's lives to continue, we must continue to conduct cutting edge research in a range of disciplines, including the underlying mathematical problems, as in this project.

The pharmaceutical sector is a huge contributor to UK PLC. The UK has a strong record of cutting edge research into the development of new treatments and drugs. AstraZeneca and GlaxoSmithKline are two of the world leading pharmaceutical companies, both having originated, and with a strong continued presence, in the UK. Further to this, other multinationals such as Pfizer, Novartis, Hoffmann-La Roche and Eisai also have a major presence in the UK. In 2007, the pharmaceutical sector employed approximately 72,000 people, and generated £8.4 billion towards the UK's gross domestic product. The continued health of this sector in the UK is important for the future health of our economy, as well as the future health of the population. This project will contribute to the development of a new area of science which will help to secure the future of this vital part of our economy.

The scope of this project is not just limited to the applications mentioned above. The mathematical methodologies developed within the project can also be applied to a whole range of problems in science and engineering. Wherever we make observations to try to understand or quantify something, whether it be observations of an oil well, or the weather, or in an MRI scanner, the methods developed in this project will aid scientists to better understand what they are observing. The breadth of impact of this area of mathematics, called the study of inverse problems, is significant.
 
Description A methodology has been developed which allows for the simplification of very complex systems of chemical reactions that occur in living cells. The methodology leads to far more accurate approximations that the methods which currently exist. Specifically, they enable this type of simplification for a much larger family of such systems than was possible previously. This work has been published.

Methodologies to parallelise and speed up methods for the quantification of uncertainty in inverse problems have been successful. Our methods show a measurable speed up compared with naive parallelisation of current methods, and also outperform other parallelised methods for certain types of problems. This work is still awaiting publication but is complete.

The combination of applying these two sets of methodologies simultaneously in order to analyse the dynamics of complex systems of chemical reactions in cells is also complete and due shortly to be submitted for publication.
Exploitation Route The project has lead to a greater understanding of multi scale systems for chemical reactions, which it seems may be applicable to many other types of dynamical systems, including but not exclusively, ODEs, PDEs and SPDEs. This is probably the most exciting output from the project.

We have also developed parallel MCMC methods which could be used to speed up inference, particularly in models with nasty multimodal behaviour, or where the posterior density is concentrated close to low-dimensional manifolds.
Sectors Aerospace

Defence and Marine

Chemicals

Energy

Environment

Manufacturing

including Industrial Biotechology

Pharmaceuticals and Medical Biotechnology

 
Description David Anderson 
Organisation University of Wisconsin-Madison
Department Mathematics
Country United States 
Sector Academic/University 
PI Contribution David Anderson and I wrote a paper together during our time at the Newton Institute in Cambridge.
Collaborator Contribution We wrote a paper on the stationary distributions of certain types of stochastic reaction networks with non-mass action kinetics.
Impact Anderson, David F., and Simon L. Cotter. "Product-form stationary distributions for deficiency zero networks with non-mass action kinetics." Bulletin of Mathematical Biology 78.12 (2016): 2390-2407.
Start Year 2016
 
Description Radek Erban 
Organisation University of Oxford
Department Oxford Centre for Human Brain Activity (OHBA)
Country United Kingdom 
Sector Academic/University 
PI Contribution Radek and I worked on a paper for analysing the error in various methods for the diffusion approximation of multi scale stochastic biochemical networks.
Collaborator Contribution Radek and I worked on a paper for analysing the error in various methods for the diffusion approximation of multi scale stochastic biochemical networks.
Impact Cotter, Simon L., and Radek Erban. "Error analysis of diffusion approximation methods for multiscale systems in reaction kinetics." SIAM Journal on Scientific Computing 38.1 (2016): B144-B163.
Start Year 2014
 
Description Yannis Kevrekidis 
Organisation Princeton University
Department Mathematics Department
Country United States 
Sector Academic/University 
PI Contribution Yannis Kevrekidis and I are continuing to work, alongside my PhD student Paul Russell, on a paper looking at Bayesian inverse problems for multi scale biochemical networks.
Collaborator Contribution Yannis Kevrekidis and I are continuing to work, alongside my PhD student Paul Russell, on a paper looking at Bayesian inverse problems for multi scale biochemical networks.
Impact Paper pending submission.
Start Year 2015
 
Description Workshop on "multi scale methods and inverse problems for biochemical networks" 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Around 30 people attended a 2 day workshop on new research on multi scale methods and inverse problems in biochemical networks.
Year(s) Of Engagement Activity 2016
URL http://www.maths.manchester.ac.uk/news-and-events/events/computationalchallenges/