Possible Research Topics

Possible research topics

Population Bioinformatics Methodology. The availability of significant numbers of high quality genomes or genotyping results and the development of new techniques for analysing this data has brought great new insights into population history. However, these new techniques all come with assumptions that affect the results and validity of the results, and give contradictory results. In some cases, results from autosomal on the one hand and MT and Y chromsome data on the other are difficult to reconcile. In many cases, the results feed into political controversy and affect group and individual identity. This project will take some public data sets and explore the impact of different assumptions, and build simulation models to see under what conditions competing methodologies can be reconciled.

Papers to explore: 10.1371/journal.pgen.1006644, 10.1093/gbe/evw046, 10.1038/ncomms3543
This topic is suitable for someone with strong mathematical or computational skills. No prior biological interest is required, though interest in the area is essential.

Microbiome. As part of the AWI-Gen project we are running a nested sub-project exploring the relationship between cardio-metabolic health and gut microbiome status. There are several sub-projects that could be explored.

Impact of diet and exercise on the gut microbiome
Use of long reads (e.g. PacBio or Nanopore data) on quality of results

De novo assemby algorithms for short read data. The huge size next generation sequence data has led to a range of approaches for assembly and clustering. Many approaches use k-mer based seeding of data -- if two sequences contain a DNA fragment of size k (typically in the range 7 to 25 depending on application), then they are considered to be candidates for being from the same region in the genome and should be clustered together. The problem with the approach is that if k is too small we get too many false posititives and hence computational costs increase, while if k is too large we miss real matches. In prior work (with older sequencing technology), we developed a heuristic called KABOOM that required two matches which we demonstrated to have better specificity and sensitivity. This project will explore the effectiveness of the approach for shorter reads.

This project is suitable for someone with very strong computational and programming background.
Papers to explore: 10.1093/bioinformatics/btr560, 10.1093/bioinformatics/btt310

Nextflow workflow for copy number variation (strong computational skills)
Copy number variants are an important source of variation in genomes. Some regions of the genome, including genes, may be duplicated (or triplicated) or missing, which may have an impact on gene expression. Tools like GenomeStrip are powerful but have strong requirements for a particular scheduler. I would like to build a robust, portable CNV detection pipeline./p>
ADME project : building reference graphs for genes is very important. There is scope for both applied work -- taking novel data that we have and building and analysing the graphs, and second developing new tools. These are MSc level projects.
Genome Wide Association Studies with complex covariates and population structure. The AWI-Gen project is investigating genomic and environmental factors in cardio-metabolic disorders in African populations. We have a large data set from a study with particpants from different countries in Africa with very different population structure and ver very different environmental conditions. We are interesed in exploring new techniques for how analysis should happen and how complex covariates and population structure should be incorporated.
- Some papers to look at are https://www.ncbi.nlm.nih.gov/pubmed/19363102, https://www.biorxiv.org/content/10.1101/439661v1.abstract, https://elifesciences.org/articles/45380