Abstracts 2008/09
MathGen Seminar Programme 2008: Michaelmas Term
Week 1: Tuesday October 14th
Jake Byrnes, WTCHG Oxford
Title: "The Application of hidden Markov models to the analysis of sequence and expression data"
Abstract: The use of probabilistic models, in particular hidden Markov models, for DNA sequence analysis has been extremely fruitful in recent years. In this talk I will describe a few HMM applications I have been developing. First, I will introduce a heterogeneous hidden Markov model for analyzing comparative tiling microarray data. The observed data for this model includes the sequence of p-values generated by the probe-wise application of any appropriate statistical test to normalized hybridization data and a vector of the inter-probe distances. Although there is bound to be some loss of power in relying on p-values, the generality of this model allows it to be configured for both simple comparisons (identifying expression or copy number differences for two lines) or for more complicated analyses (possibly involving many lines and statistical tests with numerous terms of interest).
Continuing with the theme, I will also describe a phylogenetic-HMM for identifying non-allelic gene conversion events between pairs of duplicate genes. This question is of great interest to molecular evolutionists because conversion is thought to play a significant role in duplicate gene evolution. Our model relies on a three-sequence alignment including two paralogous duplicates and one orthologous outgroup. In this case, we use the model to identify regions of reduced sequence divergence among the duplicates, suggesting a recent conversion event. The outgroup sequence is used to distinguish a reduction of divergence due to a gene conversion (observed only in the duplicates) from reduction due to the presence of a functional domain (present in all three genes).
Finally, I will briefly mention a complementary conversion identification model that relies on the patterns of polymorphism, rather than substitution, to identify more recent conversion events in a similar manner.
Amy Strange WTCHG, Oxford
Title: "Natural variation in an adaptive trait"
Abstract: The model plant Arabidopsis thaliana grows throughout the Northern hemisphere, in very diverse climates. It is able to survive in these varied environments because it has adapted accordingly. One such adaptation is the timing of flowering, in particular the vernalization response. Vernalization is the acceleration of flowering by prolonged cold. A requirement for vernalization ensures flowering occurs in the favourable conditions of spring.
Our results demonstrate that the different vernalization response phenotypes are likely to be due to variation within the same gene, giving rise to multiple independent alleles. This begins to reveal the potential of a single gene to confer micro-climate adaptation on a global scale.
Printer Friendly Version 