In the twentieth century we leant that the theory of communication is a mathematical theory. Mathematics is able to add to the value of data, for example by removing redundancy (compression) or increasing robustness (error correction). More famously mathematics can add value to data in the presence of an adversary which is my personal definition of cryptography. Cryptographers talk about properties of confidentiality, integrity, and authentication, though modern cryptography also facilitates transparency (distributed ledgers), plausible deniability (TrueCrypt), and anonymity (Tor).
Modern cryptography faces new design challenges as people demand more functionality from data. Some recent requirements include homomorphic encryption, efficient zero knowledge proofs for smart contracting, quantum resistant cryptography, and lightweight cryptography. I'll try and cover some of the motivations and methods for these.
 

As early as the 1920's Marshall suggested that firms co-locate in cities to reduce the costs of moving goods, people, and ideas. These 'forces of agglomeration' have given rise, for example, to the high tech clusters of San Francisco and Boston, and the automobile cluster in Detroit. Yet, despite its importance for city planners and industrial policy-makers, until recently there has been little success in estimating the relative importance of each Marshallian channel to the location decisions of firms.
Here we explore a burgeoning literature that aims to exploit the co-location patterns of industries in cities in order to disentangle the relationship between industry co-agglomeration and customer/supplier, labour and idea sharing. Building on previous approaches that focus on across- and between-industry estimates, we propose a network-based method to estimate the relative importance of each Marshallian channel at a meso scale. Specifically, we use a community detection technique to construct a hierarchical decomposition of the full set of industries into clusters based on co-agglomeration patterns, and show that these industry clusters exhibit distinct patterns in terms of their relative reliance on individual Marshallian channels.

The second part is to use industry relatedness, which we measure via a similar metric to co-location, to better understand the association of industrial emissions to city-industry agglomeration. Specifically, we see that industrial emissions (which are the largest source of greenhouse emissions in the US) are highly tied to certain industries, and furthermore that communities in the industry relatedness network tend to explain the tendency of particular industry clusters to produce emissions. This is important, because it limits cities' abilities to move to a greener industry basket as some cities may be more or less constrained to highly polluting industry clusters, while others have more potential for diversification away from polluting industries.

During development, cells take on specific fates to properly build tissues and organs. These cell fates are regulated by short and long range signalling mechanisms, as well as feedback on gene expression and protein activity. Despite the high conservation of these signalling pathways, we often see different cell fate outcomes in similar tissues or related species in response to similar perturbations. How these short and long range signals work to control patterning during development, and how the same network can lead to species specific responses to perturbations, is not yet understood. Exploiting the high conservation of developmental pathways, we theoretically and experimentally explore mechanisms of cell fate patterning during development of the egg laying structure (vulva) in nematode worms. We developed differential equation models of the main signalling networks (EGF/Ras, Notch and Wnt) responsible for vulval cell fate specification, and validated them using experimental data. A complex, biologically based model identified key network components for wild type patterning, and relationships that render the network more sensitive to perturbations. Analysis of a simplified model indicated that short and long range signalling play complementary roles in developmental patterning. The rich data sets produced by these models form the basis for further analysis and increase our understanding of cell fate regulation in development.

Sudden cardiac death is the most feared complication of Hypertrophic Cardiomyopathy. This inherited heart muscle disease affects 1 in 500 people. But we are poor at identifying those who really need a potentially life-saving implantable cardioverter-defibrillator. Measuring the abnormalities believed to trigger fatal ventricular arrhythmias could guide treatment. Myocardial disarray is the hallmark feature of patients who die suddenly but is currently a post mortem finding. Through recent advances, the microstructure of the myocardium can now be examined by mapping the preferential diffusion of water molecules along fibres using Diffusion Tensor Cardiac Magnetic Resonance imaging. Fractional anisotropy calculated from the diffusion tensor, quantifies the directionality of diffusion.  Here, we show that fractional anisotropy demonstrates normal myocardial architecture and provides a novel imaging biomarker of the underlying substrate in hypertrophic cardiomyopathy which relates to ventricular arrhythmia.

Tensors are higher dimensional analogues of matrices, used to record data with multiple changing variables. Interpreting tensor data requires finding multi-linear stucture that depends on the application or context. I will describe a tensor-based clustering method for multi-dimensional data. The multi-linear structure is encoded as algebraic constraints in a linear program. I apply the method to a collection of experiments measuring the response of genetically diverse breast cancer cell lines to an array of ligands. In the second part of the talk, I will discuss low-rank decompositions of tensors that arise in statistics, focusing on two graphical models with hidden variables. I describe how the implicit semi-algebraic description of the statistical models can be used to obtain a closed form expression for the maximum likelihood estimate.

The pRED Clinical Pharmacology Disease Modelling Group (CPDMG) aims to better understand the biological basis of inter-patient variability of clinical response to drugs.  Improved understanding of how our drugs drive clinical responses informs which combination dosing regimens (“right drugs”) specific patient populations (“right patients”) are most likely to benefit from. Drug evoked responses are driven by drug-molecular-target interactions that perturb target functions. These direct, "proximal effects" (typically activation and/or inhibition of protein function) propagate across the biological processes these targets participate in via “distal effects” to drive clinical responses. Clinical Systems Pharmacology approaches are used by CPDMG to predict the mechanisms by which drug combinations evoke observed clinical responses. Over the last 5 years, CPDMG has successfully applied these approaches to inform key decisions across clinical development programs. Implementation of these approaches requires: (i) integration of prior relevant biological/clinical knowledge with large clinical and “omics” datasets; (ii) application of supervised machine learning (specifically, Artificial Neural Networks (ANNs)) to transform this knowledge/data into actionable, clinically relevant, mechanistic insights.  In this presentation, key features of these approaches will be discussed by way of clinical examples.  This will provide a framework for outlining the current limitations of these approaches and how we plan to address them in the future.

Single cell RNA-Seq data is challenging to analyse due to problems like dropout and cell type identification. We present a novel clustering 
approach that applies mixture models to learn interpretable clusters from RNA-Seq data, and demonstrate how it can be applied to publicly 
available scRNA-Seq data from the mouse brain. Having inferred groupings of the cells, we can then attempt to learn networks from the data. These 
approaches are widely applicable to single cell RNA-Seq datasets where  there is a need to identify and characterise sub-populations of cells.