Gene Network Analysis in Drug Discovery and Development

In the post-genomic era, network based analysis enables us to reveal biological systems not at individual component level but at a system-wide level. Gene regulatory networks, which are inferred from microarray data, are a method that can provide valuable information about the regulation of gene expression in cells. This technology is also applied to the drug discovery and development (DDD) field to identify drug target genes and reveal their modes of action. The talk overviews what a gene network is, how it is inferred and how it is used, especially from the point of view of DDD. Finally, I will show our work that identified a new target gene of a well-known compound in endothelial cells using gene network analysis.

Opportunities around Transactional Medical Records with Application to Chronic Condition Management

Data that is recorded as part of the operation of healthcare per se is a fantastic resource to apply for quality improvement in healthcare delivery. Prescribing and laboratory test results are two of the best elements to use as these are objective and functional (as compared to, say, the notation of a problem or symptom). These data are recorded as acts at points in time, however, which leaves the problem of inferring the bigger picture. For example, one instance of an antihypertenisive prescription can imply hypertension (although not always reliably), and a clinical decision to pursue lifelong pharmacological management (unless heroic lifestyle changes are made by the patient).

The talk overviews NIHI research on analysis of electronic health record (EHR) data, particularly recent examination of general practice records around treatment of hypertension. We will show how the findings highlight problems and opportunities around medication adherence, and look at the broader issues and applications for inference from EHR data in chronic condition management.

The Analysis of Unclassified Variants in Clinical Molecular Genetics

The increasing availability of high throughput sequencing technologies and the commensurate increase in clinical genetic testing presents new challenges and opportunities. In the context of clinical diagnostic labs, unclassified genetic variants with unknown effects often present themselves. One of the major challenges today in clinical genetics is to predict the pathological consequences of variants and nonsynonymous single nucleotide polymorphisms (nsSNP) observed in patients under fixed time and budget constraints. I will discuss some of the bioinformatic approaches to addressing this problem and present a new decision-support tool for diagnostics.

Handling missing data from high-throughput protein abundance experiments

I will describe a two-phase experiment designed to compare the protein abundances of the inner and outer left ventricle walls of healthy and diabetic rats’ hearts. In the first phase experiment, rats are assigned to healthy or diabetic treatment groups, while in the second phase the biological material is processed in a laboratory-based experiment and protein abundances are measured using MudPIT (Multi-dimensional Protein Identification Technology) coupled with iTRAQTM (isobaric Tags for Relative and Absolute Quantitation). The experiment was designed assuming every protein species be detected in each protein mixture sample. However, this is not the case in practice; some proteins, although present, are simply never observed in some MudPIT runs. I discuss the consequences of missing runs on the design and analysis of this experiment, and the implications for designing future MudPIT-iTRAQ experiments.

Experiences fitting graphical models with the graphical lasso

The lasso is a familiar technique for regularization in regression problems, and recent advances have extended the methodology to regularization of inverse covariance matrices. The talk will describe the rationale behind lasso and related techniques such as adaptive lasso and SCAD. We will then apply the lasso to some moderately sized microarray data sets and explore what insights can be gained from this type of modeling.

Gene Network Inference and Application

My work focuses on inferring regulatory networks from gene expression data, and applying them to problems in human biology. In this talk, I will explain what a gene network is and how it might be useful; describe some new tools and ideas developed to make better use of gene expression data; and show some of the most exciting results we have found in different types of cells. In particular, I will show some examples of using network inference techniques to study transcription factor interactions in endothelial cells, and explain a new hypothesis we have proposed to account for interactions up- and downstream of particular transcription factors. Finally, I will talk about our work on malignant melanoma, and show some results from combining a new gene expression dataset with published clinical data from patient tumours to identify previously unknown master regulators associated with differences in survival of melanoma patients.

The data explosion: we have met the enemy and he is us

In his poem “The Rock” T. S. Eliot writes: “Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information?” Had he been writing at the start of the 21st Century, I think he may also have added: “Where is the information we have lost in data?”

In science and in government, we are now faced with mountains of data, growing at a geometric rate, and a similarly daunting mass of published papers and reports that we are expected to find and consult. The amount of raw storage that science requires is doubling every 18 months on average, and even faster in some fields, such as bioinformatics. Who is going to provide all that storage? Where will it reside? Similarly, f every scientific article were to be kept physically in the national library, then the size of the building would need to double every two years. How can we possibly keep up?

We have become very proficient at getting information online, but we remain ill-equipped to find the information we need precisely, or to understand its content effectively. As containers of important knowledge, mot of our digital resources are very inefficient to explore or examine.

We need to get far more efficient at describing what we know, what we do, and how it might be relevant to others; and conversely, at searching to find what others have done and said that might be relevant to us. Alternatively, we can just continue to create datasets and reports that nobody else can use because they cannot find them or cannot figure out if they are appropriate to use!

This talk examines new approaches to capturing, representing, visualising and conveying important facets of data content in very large data collections.

Missing heritability in genome-wide association studies

There is currently considerable debate about the role of genes in complex disorders. Genome-wide association studies have greatly enhanced out understanding of the genetic basis of many common human diseases. However, the current analytical method only account for around 25% of the variability. Bruce will discuss some of the current controversy, drawing on his own findings.

Bruce Weir directs the Coordinating Centers for GENEVA and GARNET, two collections of GWAS studies funded by the National Human Genome Institute in the USA.

Drink and nibbles will be served from 4:15. All welcome

Finding needles in a haystack: identification of significant molecular changes in influenza viruses

The influenza virus is highly variable. Genetic changes in the virus may have important effects on viral properties. For example, some changes allow the virus to escape from vaccine protection, thereby necessitating frequent updates of the vaccine formulation. However, it appears that most viral genetic variation has no significant effect on viral properties. The challenge is to find effective approaches for identifying the few "important" changes in a background of a large amount of "noise."

Experimental approaches offer some success. However, the practicalities of experiments mean that only a small domain of variation can be explored, and thus the scope of inferences is limited. We have been developing bioinformatics approaches to the problem of finding the needles in the haystack. A potential benefit of bioinformatics approaches is broadly applicable inferences, leading to deeper insights into significant changes in the influenza virus.

There are significant statistical obstacles to these bioinformatics developments. Fundamental to our developments is a precise understanding of the way in which the influenza virus evolves. If we know how the virus evolves, we can assess the relative likelihood that a change is important or insignificant. The influenza virus evolves in complex and unusual ways, necessitating novel statistical models, which we have been developing. Further, in the influenza virus, it appears that important molecular changes rarely operate in isolation. More likely, two co-ordinated changes are needed to achieve a different-and-fitter virus. Finding these co-ordinated changes in the highly variable influenza virus requires new developments.

This talk is intended as a high-level view of our statistical, computational and experimental work toward proposing experimentally testable hypotheses of significant changes in the influenza virus. The talk will not expect any knowledge of influenza virus genetics or evolution.

Quality control issues in next generation sequencing experiments

NGS experiments are creating vast amounts of data requiring significant computing resources to effectively process it. Even thought the information derived from these sequencing technologies is of a concise format, the quantity of information is challenging to handle. There also appear to be technical issues that need to be addressed similar to microarray technology in its outset. This talk will discuss quality control, visualization and data exploration of next generation sequencing data.