Visit my current homepage.

"There are no mathematical equations in On the Origin of Species. A good thing too, you might think, and it is undoubtedly true that Darwin's clear and flowing narrative style helped ensure the popularity of his writings. Modern research in evolutionary biology can make for less easy reading. Much of it concerns the development of an expanding arsenal of mathematical and statistical techniques, necessary to do battle with the relentless onslaught of gene and genome sequences. Of course, the discrete, ordered nature of genetic information and the stochastic character of Mendelian inheritance have naturally lent themselves to numerical analysis. Consequently, the mathematical foundations of evolutionary genetics have, somewhat unusually for biology, tended to precede the data to which they are applied. The Genetical Theory of Natural Selection by R. A. Fisher, published only fifty years after Darwin's death, is full of equations." - Oliver G. Pybus, "Model selection and the molecular clock", PLoS Biology, May 2006.

Probabilistic models of evolution

With the availability of massive amounts of genomic sequence data, it has become possible to start piecing together various aspects of the history of the organisms on this planet. By using ever more sophisticated probabilistic models, we are starting to get to grips with the details of the process by which natural selection has shaped, and continues to shape, the evolution of genes and genomes. For instance, we are able to make remarkable inferences about the selective pressures acting on individual amino acid sites in proteins as they evolve in a group of related species. This is of particular importance when studying pathogens, where positive selective pressure is often a result of an "arms race" with the immune system of the host species and hence provides clues that can enable biologists to understand the function of important genes and, hopefully, develop ways of countering the pathogens that use them.

Background

I completed my PhD in 2002 at the Cambridge University Engineering Department, where I was a member of the Speech, Vision and Robotics group (now called the Machine Intelligence group). My thesis dealt with automatic design of spoken dialogue systems, including quantitative modelling of speech recognition systems along with their users, quantitative simulation of human-computer dialogue, and the application of reinforcement learning to strategy design in spoken dialogue systems. Prior to this I obtained the B.Eng and M.Eng degrees in Electronic Engineering (specialising in speech recognition) at the University of Stellenbosch.

After a year and a half as lecturer in Computer Science at the University of the Western Cape, I started working as a postdoctoral research fellow first at the South African National Bioinformatics Institute (SANBI) and then at the computational biology group at the University of Cape Town, where I have been since January 2005. Concurrently with this, I am currently a visiting researcher at Stanford University (February-July 2007).

Recent publications (follow the links for more information and supplementary materials):

A Model of Directional Selection Applied to the Evolution of Drug Resistance in HIV-1, C. Seoighe, F. Ketwaroo, V. Pillay, K. Scheffler, N. Wood, R. Duffet, M. Zvelebil, N. Martinson, J. McIntyre, L. Morris and W. Hide, Molecular Biology and Evolution, 2007: We implemented a model of directional selection to detect instances where specific amino acids are favoured at a site in a protein. The model is particularly suited for detecting signals of convergent evolution, as one expects to find when comparing intra-patient HIV data across many patients.

Maximum Likelihood Inference of Imprinting and Allele-specific Expression from EST Data, C. Seoighe, V. Nembaware and K. Scheffler, Bioinformatics, 2006: We built probabilistic models of imprinting and allele-specific expression. Application to data from dbEST resulted in a list of candidates (highly significant enrichment for genes known to be imprinted) that can be investigated experimentally.

Robust inference of positive selection from recombining coding sequences, K. Scheffler, D.P. Martin and C. Seoighe, Bioinformatics 2006: Traditional methods for detecting positive selection suffer from high levels of false positives when recombination is present. In this paper we present a method that restores good statistical properties. From the Faculty of 1000 review: "This paper provides a solution to one of the most important problems in evolutionary bioinformatics".

A Bayesian Model Comparison Approach to Inferring Positive Selection, K. Scheffler and C. Seoighe, Molecular Biology and Evolution, 2005: This paper describes a simulation study comparing maximum likelihood (using PAML) and Bayesian (using MrBayes, with a novel model comparison step) approaches for detecting positive selection. The new BEB (Bayesian empirical Bayes) analysis in PAML clearly outperforms the older NEB (naive empirical Bayes) analysis, and, under most circumstances, gives similar results to the more computationally intensive Bayesian analysis. Bayesian analysis is recommended when your trees are either very short (clear improvement in power at the same false positive rate) or saturated (you won't get informative results, but at least you won't get misleading results).

Very low power to detect asymmetric divergence of duplicated genes, C. Seoighe, K. Scheffler, Proc. RECOMB 2005 International Workshop on Comparative Genomics. A. McLysaght and D. Huson (Eds). Springer, Heidelberg: We found that previously published methods for detecting asymmetric divergence of duplicated genes (where the two copies evolve at different rates) have serious sources of bias. Removing the problem leaves a method with very low power and consequently estimates of the magnitude of this effect in the evolution of duplicate genes should not be trusted.

Detecting molecular evidence of positive Darwinian selection, K. Scheffler and C. Seoighe, in ``Information Processing And Living Systems'', V. Bajic and Tan Tin Wee (eds), World Scientific Publishing, 2005: A book chapter reviewing phylogenetic and other methods for detecting positive selection.

Old homepages

At SANBI.
At UWC Computer Science.
At Cambridge University (Machine Intelligence group).