Our paper is the first ever to describe a formal methodology to assess how
well a number of mathematical models describe a specific biological
mechanism. This ability to rank plausible models based on experimental
evidence is of enormous importance to life scientists as it provides a
formal, objective, and rational means with which to study the many possible
hypotheses that could describe the mechanisms being studied.
As an example, cancer biologists have been studying the ERK pathway, also
known as the p42/p44 MAP kinase pathway, for several decades, and there are
many working hypotheses regarding the structure and dynamics of this
signalling cascade. Many mathematical models have previously been developed
to aid our understanding of ERK signalling, but there has been little
attention paid to formally assessing the biological credibility of these
"My research in the Inference
Research Group at Glasgow has been focused on
the development of computational techniques
at the life sciences
The work in our paper enables this assessment to be rigorously carried out
and has subsequently been successful in identifying the role of BRaf in ERK
signalling by the interplay of mathematical modelling, evidence ranking,
and subsequent guided experimentation. It is for this reason that the
scientific community recognizes the foundational importance of this work
and hence the high level of citation of this paper.
Because the methodology described in the paper has a solid statistical
foundation, in that it embeds mathematical models within the Bayesian
inferential framework, it has attracted a great deal of interest from the
computing science and mathematical statistics research communities.
Our development and subsequent use of advanced statistical methods based on
thermodynamic integration is at the very cutting edge of statistical and
computational methodology. In other words, advanced computational
statistics is being used to drive forward scientific enquiry and discovery.
Does it describe a new discovery, methodology, or
synthesis of knowledge?
To our knowledge, and to that of the journal editor and referees, our paper
is the first to describe the novel synthesis of mathematical modelling and
Bayesian statistical inference in providing a novel methodology, which is
the cornerstone to support scientific enquiry.
Would you summarize the significance of your paper in
Advances in the availability of affordable computing power have made it
possible to build mathematical models of biological systems which control,
for example, the beating of a heart, the genesis of serious diseases, and
possible new drug therapies. The main issue about these models is that
there may well be a number of different model descriptions which all are
capable of simulating the behavior of the biological system being studied.
However, when using these models to make predictions, such as what impact a
certain drug will have on the control of malignant tumors, they may suggest
wildly different outcomes. It is for this reason that it is of the utmost
importance that a formal, objective, and rational method to assess the
validity of the range of models considered is available.
Our paper describes such a methodology that employs advanced mathematical,
statistical, and computing techniques. The power of our method has already
been successfully demonstrated by a number of life scientists working in
the field of cancer biology.
How did you become involved in this research, and were
there any problems along the way?
My research in the Inference Research Group at Glasgow has been focused
on the development of computational techniques at the life sciences
interface. I was especially drawn to the Bayesian formalism for
scientific inference and argued that it was, in effect, a formal
representation of the scientific method itself.
The major challenges that had to be addressed were the computational effort
required for estimating the evidence in support of a mathematical model
when employing Markov Chain Monte Carlo (MCMC) sampling methods. A whole
research program in developing efficient MCMC methods has emerged from this
initial work. An additional challenge was to make the multidisciplinary
collaboration across the mathematical and biological disciplines effective.
Where do you see your research leading in the
I am interested in continued research at the mathematical and life sciences
interface as there are many opportunities there to make an impact of real
scientific and societal significance.
Do you foresee any social or political implications for
There are very clear implications that this work will lead to improvements
in healthcare and pharmaceutical research.
Dr. Vladislav Vyshemirsky
Department of Computing Science
University of Glasgow
View a video lecture by Vladislav Vyshemirsky on the
subject of this paper.