The projects below are either potential or current/completed projects by Masters and Undergraduate students in my lab. These or similar projects, or extensions of completed projects, may be potential summer work, senior honours projects; more in-depth or advanced versions may be potential Master’s projects.
Experimental exploration of pathways in the stress response of yeast
Yeast stress response consists of several intercalated pathways. This project will explore gene and protein expression involved in yeast stress response. Techniques applied may include: yeast transformation, creation of flourescent protein-fusions, quantitative PCR, flow cytometry. Student interest will guide the exact question explored; this could range from studying the effect of experimental evolution under stress conditions on the expression of a stress response protein, to detailed examination of a particular pathway, to working with synthetically engineering circuits.
Revealing networks of gene regulatory interactions based on analysis of gene microarray data
All our cells (with a few exceptions) have the same DNA, and yet cells in different tissues, and over developmental time, can have within them widely different proteins. This happens through differential gene regulation. Using gene expression data from microarrays, we can apply functional network inference algorithms to attempt to reveal these gene regulatory networks. This project will involve application of Bayesian network inference algorithms to publicly available gene expression data. There is much current work in bioinformatics and computational biology on the analysis of gene regulatory networks; this project will contribute to this work by exploring ways for getting the most out the computational tools, using techniques such as influence scores and model averaging. The expression data chosen will depend on the student’s interest, and can range from topics like environmental stress in yeast to breast cancer.
Revealing ecological networks based on analysis of ecosystem data
Organisms and habitats interact in a complex web of interactions within an ecosystem. We have recently found that Bayesian network inference algorithms are capable of revealing these interactions based on species count and habitat data. This project will further explore this new application of Bayesian networks by applying them to ecological data. There are a number of data sets which may be available, such as avian, marine, or vegetation systems; student interest will direct the system studied.
Exploring measures of social association using computer simulation
A difficulty in evaluating experimental measures of social association in animals is that the true association pattern is unknown. This project explores measures of social association by building computer models of assorting animals, and then simulating different methods of data sampling. Because we know the rules governing the animals’ behaviour from the computer model, it is possible to evaluate the effectiveness of the measurements.
Looking for the best method to find a Bayesian network to describe different types of biological networks
We have used Bayesian networks to analyse a number of different types of biological networks, genetic, neuronal, and ecological. While the algorithms are useful in all cases, the type and amount of data analysed can influence the details of its implementation. For example, when analysing neural data, we find that the algorithm tends to “agree” on a solution—it will find the same network in independent runs of the heuristic search; but when analysing ecological data, different searches may find different networks, with only a few commonalities consistently recovered. This project will explore what leads to these differences in ability to “agree” on a network—whether it is a feature of the system or of the amount of data available, or both. The project will involve application of Bayesian network inference algorithms to both real and simulated data from multiple types of biological networks.
Current/Completed Masters Projects
Revealing plankton and herring food webs in the Irish Sea (MRes 2008, Emily King)
Plankton is the main food source of herring, an economically important species. Using many decades of data of herring and plankton abundance, this project explores networks among plankton species and with herring in the Irish Sea. The networks will be used to create predictive models which may be useful for management of herring fisheries, as well as providing new insight to pelagic ecosystems.
Current/Completed Undergraduate Projects
Bioinformatics and Computational Biology
Comparison of Bayesian network algorithms’ ability to recover networks related to the cMYC proto-oncogene from human lymphocyte data (SH 2008-9, Ryan Gallagher)
This project compares performance of static and dynamic Bayesian network inference algorithms on human microarray gene expression data, for recovering networks related to cMYC, with previously published results.
Habitat and species interactions among bird populations in the Peak District National Park (Summer 2006, Isobel Milns)
Understanding interactions within ecological networks can help reveal keys to ecosystem stability/fragility, provide knowledge for management, or help refine models of species distribution change. Current models of ecological networks require detailed measurement—or assumption—of parameters such as birth and death rate, encounter frequency, and predation success that become unfeasible as the number of species in the ecosystem increases. Bayesian network inference algorithms, previously applied to analyse gene regulatory and neural information flow networks, have the potential to reveal ecological networks based upon species and habitat abundance alone. This project tests the algorithm’s performance and applicability on observational data of avian communities and habitat in the Peak District National Park.
An investigation and evaluation of neural prosthetics (SH Reading 2008, Katherine Culshaw)
New research in neural function and human-machine interfaces have produced the possibility of neural-controlled prosthetic replacements for missing limbs or senses. This reading project investigates the development of such prosthetics, the algorithms used to control them, and evaluates the potential success of neural prosthetics as a whole and of different subtypes.
Revealing networks of neural information flow based on analysis of neural electrophysiology recordings (SH 2007, Joshua Peck)
Behaviour and perception depend on activity within the neural connections in the brain. Understanding the networks of neural information flow and how they change over time and for different behaviour may help us understand how the brain works. We have recently shown that Bayesian network inference algorithms are capable of revealing these neural information flow networks when applied to electrophysiology recordings. This project involves application of these algorithms to multi-unit electrophysiology data collected from the auditory regions of female zebra finches; among other things, the student will explore whether or not the networks differ when looking across different time scales, indicating multiple levels of information coding in the brain.
Exploration and Modification of a Teaching Strain of Yeast (SH 2008-9, Kathryn Munro)
First year Molecular Biology students explore the mystery of the “Red and White Yeast” in a laboratory practical at the end of the year. This senior honours project further explores this yeast strain, and engineers new strain(s) for use in future years.
Characterisation of single cell variation in protein expression of stress induced proteins in yeast (Summer 2007, Christina Müller)
Gene and protein expression are usually measured as population means, which does not provide information on individual variation in across cells. However, many computational analyses need to make assumptions about the distribution of such individual variation. This project measures and characterises the individual variation in the expression of yeast stress reponse proteins. GFP-fusions to stress response proteins of interest will be made. Then measurements will be taken in a flow cytometer, where fluorescence level of a cell corresponds to its protein expression level. The distribution across individual cell’s will be fit to standard statistical distributions, and the response of the expression over time and level of stress will also be examined.
Creation of genetically modified yeast with fluorescent fusion proteins and titratable promoters (Summer 2006, Christina Müller)
Amount of fluorescence can be measured quantitatively on a cell-by-cell basis using a fluorescence flow cytometer. This technique can be used to quantify the level of protein expression of proteins of interest, if they are fused to fluorescent proteins (for example, GFP—green fluorescent protein, RFP—red fluorescent protein). This project involves the creation of genetically modified yeast (Saccharomyces cerevisiae) strains having fluorescent fusion proteins to putative regulatory and target genes involved in yeast stress response. Plasmids containing the desired genetic material must first be transformed into E. coli for production of large quantities of the DNA, and then the plasmids must be extracted, the relevant DNA copied using PCR, and transformed into yeast. Finally, putative regulators will have titratable promoters which will be varied to induce different levels of expression of the gene, and thus the protein. Success of the modifications will be observed using flow cytometry and/or fluorescent microscopy.