Past Mathematical Biology and Ecology Seminar

Current medical practice is driven by the experience of clinicians, by the difficulties of integrating enormous amounts of complex and heterogeneous static and dynamic data and by clinical guidelines designed for the “average” patient. In this talk, I will describe some of my research on developing novel, specially-crafted machine learning theories, methods and systems aimed at extracting actionable intelligence from the wide variety of information that is becoming available (in electronic health records and elsewhere) and enabling every aspect of medical care to be personalized to the patient at hand. Because of the unique and complex characteristics of medical data and medical questions, many familiar machine-learning methods are inadequate.  My work therefore develops and applies novel machine learning theory and methods to construct risk scores, early warning systems and clinical decision support systems for screening and diagnosis and for prognosis and treatment.  This work achieves enormous improvements over current clinical practice and over existing state-of-the-art machine learning methods.  By design, these systems are easily interpretable and so allow clinicians to extract from data the necessary knowledge and representations to derive data-driven medical epistemology and to permit easy adoption in hospitals and clinical practice. My team has collaborated with researchers and clinicians in oncology, emergency care, cardiology, transplantation, internal medicine, etc. You can find more information about our past research in this area at: http://medianetlab.ee.ucla.edu/MedAdvance.

Building on advancements in computer vision we now have an array of visual tracking methods that allow the reliable estimation of cellular motion in high-throughput settings as well as more complex biological specimens. In many cases the underlying assumptions of these methods are still not well defined and result in failures when analysing large scale experiments.

Using organotypic co-culture systems we can now mimic more physiologically relevant microenvironments in vitro.  The robust analysis of cellular dynamics in such complex biological systems remains an open challenge. I will attempt to outline some of these challenges and provide some very preliminary results on analysing more complex cellular behaviours.

In the first part of my presentation, I will briefly summarize a dynamic view of the cell cycle created in collaboration with Prof John Tyson over the past 25 years. 
In our view, the decisions a cell must make during DNA synthesis and mitosis are controlled by bistable switches, which provide abrupt and irreversible transition 
between successive cell cycle phases. In addition, bistability provides the foundation for 'checkpoints' that can stop cell proliferation if problems arise 
(e.g., DNA damage by UV irradiation). In the second part of my talk, I will highlight a few representative examples from our ongoing BBSRC Strategic LoLa grant 
(http://cellcycle.org.uk/) in which we are testing the predictions of our theoretical ideas in human cells in collaboration with four experimental groups.

Proteomics is seen as the next logical step after genomics to understand life processes at the molecular level. With increasing capabilities of modern mass spectrometers the deep proteome (>8000 proteins detected) has become widely accessible, only to be replaced recently by the "Ultra-deep proteome" with ~14000 proteins detected in a single cell line. Furthermore, new data search algorithms and sample preparation methods allow not only to achieve comprehensive sequence coverage for the majority of proteins, but also to detect protein variations and single amino acid polymorphisms in proteins, further linking genomic variation to protein phenotypes. The combination of genomic and proteomic information on individual (patient) level could mark the beginning of the "Post-Proteomic Era".

Please register via https://www.eventbrite.co.uk/e/qbiox-colloquium-trinity-term-2017-ticke…

The ability to study the transcriptome, proteome – and other aspects – of many individual cells represents one of the most important technical breakthroughs and tools in biology and medical science of the past few years. They are revolutionising study of biological systems and human disease, enabling for example: hypothesis-free identification of rare pathogenic (or protective) cell subsets in chronic diseases, routine monitoring of patient immune phenotypes and direct discovery of mole cular targets in rare cell populations. In parallel, new computational and analytical approaches are being intensively developed to analyse the vast data sets generated by these technologies. However, there is still a huge gap between our ability to generate the data, analyse their technical soundness and actually interpret them. The QBIOX network may provide for a unique opportunity to complement recent investments in Oxford technical capabilities in single-cell technologies with the development of revolutionary, visionary ways of interpreting the data that would help Oxford researchers to compete as leaders in this field.

Please register via https://www.eventbrite.co.uk/e/qbiox-colloquium-trinity-term-2017-ticke…

Biomedical research and clinical practice rely on complex and multimodality

datasets for the characterisation of human organs in health and disease. In

computational biomedicine, we often argue that multiscale computational

models are and will be increasingly required as tools for data integration,

for probing the established knowledge of physiological systems, and for

predictions of the effects of therapies and disease. But what has

computational biomedicine delivered so far? This presentation will describe

successes, failures and future directions of computational models in

cardiac research from basic to translational science.

All data intelligence processes are designed for processing a finite amount of data within a time period. In practice, they all encounter
some difficulties, such as the lack of adequate techniques for extracting meaningful information from raw data; incomplete, incorrect 
or noisy data; biases encoded in computer algorithms or biases of human analysts; lack of computational resources or human resources; urgency in 
making a decision; and so on. While there is a great enthusiasm to develop automated data intelligence processes, it is also known that
many of such processes may suffer from the phenomenon of data processing inequality, which places a fundamental doubt on the credibility of these 
processes. In this talk, the speaker will discuss the recent development of an information-theoretic measure (by Chen and Golan) for optimizing 
the cost-benefit ratio of a data intelligence process, and will illustrate its applicability using examples of data analysis and 
visualization processes including some in bioinformatics.

The management of natural resources, from fisheries and climate change to gut bacteria colonies, all require the development of ecological models that represent the full spectrum of population interactions, from competition, through mixotrophy and mutualism, to predation.

Mixotrophic plankton, that both photosynthesise and eat other plankton, underpin all marine food webs and help regulate climate by facilitating gas exchange between the ocean and atmosphere. We show the recent discovery that their feeding preferences change with increasing temperature implies climate change could dramatically alter the structure of marine food webs.

We describe a theoretical framework that reveals the key role of mixotrophy in facilitating transitions between trophic interactions. Mixotrophy smoothly and stably links competition to predation, and extends this linkage to include mutualism in both facultative and obligate forms. Such smooth stable transitions further allow the development of eco-evolutionary theory at the population level through quantitative trait modelling.

Regenerative medicine offers great hope in curing many currently untreatable diseases. Tissue engineering and stem cell therapy are the two main components of regenerative medicine. In this talk, I will discuss how engineering can make contributions to this highly interdisciplinary field, including biomaterials as 3D scaffolds, bioreactor design, and stem cell bioprocessing.

We adopt the paradigm of understanding how the heart develops during pregnancy as a first principal to inform on adult heart repair and regeneration. Our target for cell-based repair is the epicardium and epicardium-derived cells (EPDCs) which line the outside of the forming heart and contribute vascular endothelial and smooth muscle cells to the coronary vasculature, interstitial fibroblasts and cardiomyocytes. The epicardium can also act as a source of signals to condition the growth of the underlying embryonic heart muscle. In the adult heart, whilst the epicardium is retained, it is effectively quiescent. We have sought to extrapolate the developmental potential of the epicardium to the adult heart following injury by stimulating dormant epicardial cells to give rise to new muscle and vasculature. In parallel, we seek to modulate the local environment into which the new cells emerge: a cytotoxic mixture of inflammation and fibrosis which prevents cell engraftment and integration with survived heart tissue. To this end we manipulate the lymphatic vessels in the heart given that, elsewhere in the body, the lymphatics survey the immune system and modulate inflammation at peripheral injury sites. We recently described the development of the cardiac lymphatic vasculature and revealed in the adult heart that they undergo increased vessel sprouting (lymphangiogenesis) in response to injury, to improve function, remodelling and fibrosis. We are currently investigating whether increased lymphangiogenesis functions to clear immune cells and constrain the reparative response for optimal healing.

Oxford Nanopore Technologies aim to enable the analysis of any living thing, by any person, in any environment. The world's first and only nanopore DNA
sequencer, the MinION is a portable, real time, long-read, low cost device that has been designed to bring easy biological analyses to anyone, whether in
scientific research, education or a range of real world applications such as disease/pathogen surveillance, environmental monitoring, food chain
surveillance, self-quantification or even microgravity biology. Gordon will talk the about the technology, applications and future direction.
Stuart will talk about the nanopore signal, computational methods and informatics challenges associated with reading DNA directly.

Atomistic Molecular Dynamics is a well established biomolecular modelling tool that uses the wealth of information available in the Protein Data Bank (PDB). However, biophysical techniques that provide structural information at the mesoscale, such as cryo-electron microscopy and 3D tomography, are now sufficiently mature that they merit their own online repository called the EMDataBank (EMDB). We have developed a continuum mechanics description of proteins which uses this new experimental data as input to the simulations, and which we are developing into a software tool for use by the biomolecular science community. The model is a Finite Element algorithm which we have generalised to include the thermal fluctuations that drive protein conformational changes, and which is therefore known as Fluctuating Finite Element Analysis (FFEA) [1].

We will explain the physical principles underlying FFEA and provide a practical overview of how a typical FFEA simulation is set up and executed. We will then demonstrate how FFEA can be used to model flexible biomolecular complexes from EM and other structural data using our simulations of the molecular motors and protein self-assembly as illustrative examples. We then speculate how FFEA might be integrated with atomistic models to provide a multi-scale description of biomolecular structure and dynamics.

1. Oliver R., Read D. J., Harlen O. G. & Harris S. A. “A Stochastic finite element model for the dynamics of globular macromolecules”, (2013) J. Comp. Phys. 239, 147-165.

A holy grail of nano-technology is to create truly complex, multi-component structures by self assembly.
 

Most self-assembly has focused on the creation of `structural complexity'. In my talk, I will discuss `Addressable Complexity': the creation of structures that contain hundreds or thousands of
distinct building blocks that all have to find their place in a 3D structure.

If the abundances of the constituent molecules of a biochemical reaction system  are sufficiently high then their concentrations are typically modelled by a coupled set of ordinary differential equations (ODEs).  If, however, the abundances are low then the standard deterministic models do not provide a good representation of the behaviour of the system and stochastic models are used.  In this talk, I will first introduce both the stochastic and deterministic models.  I will then provide theorems that allow us to determine the qualitative behaviour of the underlying mathematical models from easily checked properties of the associated reaction network.  I will present results pertaining to so-called ``complex-balanced'' models and those satisfying ``absolute concentration robustness'' (ACR).  In particular, I will show how  ACR models, which are stable when modelled deterministically, necessarily undergo an extinction event in the stochastic setting.  I will then characterise the behaviour of these models prior to extinction.

Transmural propagation is a little studied feature of mammalian electrophysiology, this talk reviews our experimental work using different optical techniques to characterise this mode
of conduction under physiological and pathophysiological conditions.

We will discuss how the analysis of a stochastic mean-field model for
synaptic activity can be used to reconstruct some parameters about
neuronal networks.  The method is based on a non-standard analysis of the
Fokker-Planck equation and the asymptotic computation of the spectrum for
the nonself-adjoint operator. Applications concern Up- and Down- states
and bursting activity in neuronal networks.

Simulating the long time and length scales associated with DNA self-assembly
and DNA nanotechnology is not currently feasible with models at an atomic level
of detail. We, therefore, developed oxDNA a coarse-grained representation of
DNA that aims to capture the fundamental structural, thermodynamic and
mechanical properties of double-stranded and single-stranded DNA, which we have
subsequently applied to study a wide variety of DNA biophysical properties and
DNA nanotechnological systems.

Ductal carcinoma is one of the most common cancers among women, and the main cause of death is the formation of metastases. The development of metastases is caused by cancer cells that migrate from the primary tumour site (the mammary duct) through the blood vessels and extravasating they initiate metastasis. In my talk, I present a multi-compartment mathematical model which mimics the dynamics of tumoural cells in the mammary duct, in the circulatory system and in the bone. Using a branching process approach, the model describes the relation between the survival times and the four markers mainly involved in metastatic breast cancer (EPCAM, CD47, CD44 and MET). In particular, the model takes into account the gene expression profile of circulating tumour cells to predict personalised survival probability. Gene expression data of metastatic breast cancer have been used to validate the model. The administration of drugs as bisphosphonates is also included in order to analyse the dynamic changes induced by the therapy.

Stochastic and deterministic processes are merged together to describe cancer progression and obtain personalised survival analysis based on the gene expression levels of each patient. The main aim of the talk is showing that Mathematics can have a strong impact in speeding cancer research, predicting survival probability and selecting the best cancer treatment. 

The first half of the lecture will begin by reviewing what is known about the
neural representation of faces in the primate visual system. How does the
visual system represent the spatial structure of faces, facial identity and
expression? We then discuss how depression is associated with negative
cognitive biases in the recognition of facial expression, whereby depressed
people interpret facial expressions more negatively. The second half of the
lecture presents computer simulations aimed at understanding how these facial
representations may develop through visual experience. We show how neural
representations of expression are linked to particular spatial relationships
between facial features. Building on this, we show how the synaptic connections
in the model may be rewired by visual training to eliminate the negative
cognitive biases seen in depression.

It is well known that stochasticity can play a fundamental role in 
various biochemical processes, such as cell regulatory networks and 
enzyme cascades. Isothermal, well-mixed systems can be adequately 
modeled by Markov processes and, for such systems, methods such as 
Gillespie's algorithm are typically employed. While such schemes are 
easy to implement and are exact, the computational cost of simulating 
such systems can become prohibitive as the frequency of the reaction 
events increases. This has motivated numerous coarse grained schemes, 
where the ``fast'' reactions are approximated either using Langevin 
dynamics or deterministically.  While such approaches provide a good 
approximation for systems where all reactants are present in large 
concentrations,  the approximation breaks down when the fast chemical 
species exist in small concentrations,  giving rise to significant 
errors in the simulation.  This is particularly problematic when using 
such methods to compute statistics of extinction times for chemical 
species, as well as computing observables of cell cycle models.  In this 
talk, we present a hybrid scheme for simulating well-mixed stochastic 
kinetics, using Gillepsie--type dynamics to simulate the network in 
regions of low reactant concentration, and chemical langevin dynamics 
when the concentrations of all species is large.  These two regimes are 
coupled via an intermediate region in which a ``blended'' jump-diffusion 
model is introduced.  Examples of gene regulatory networks involving 
reactions occurring at multiple scales, as well as a cell-cycle model 
are simulated, using the exact and hybrid scheme, and compared, both in 
terms weak error, as well as computational cost.

This is joint work with A. Duncan (Imperial) and R. Erban (Oxford)