Applying statistical models to provide novel insights into the ageing process
Sasdekumar Loganathan, Atefeh Taherian Fard, Ameya Kulkarni and Jessica Mar
Australian Institute for Bioengineering and Nanotechnology (AIBN), University of Queensland, Queensland, Australia
Ageing is a complex process where the combined effects of environmental and genetic factors make it challenging to isolate specific regulators of ageing that come from the genome. Moreover, both the variation in ageing between individuals and the variation between different tissues make the task of identifying regulators even more challenging. The use of chronological age as a predictor of an ageing phenotype is one example of how this metric does not translate accurately to everyone in a human population.
Standard models in bioinformatics operate under the assumption that differences between two phenotypic groups are captured by mean differences only. However, changes in gene expression variability have shown to contain regulatory information too. This project investigates the variability of gene expression and its relationship to the regulation of aging by modelling gene expression profiles using mixture models. A mixture model describes heterogeneity within a population through the representation of multiple modes or sub-populations in the data. We applied mixture models to a tissue-specific gene expression dataset, Genotype-Tissue Expression (GTEx), derived from multiple tissues from donors without chronic diseases. Though this dataset was not designed with the original intention to study ageing, it provides an excellent opportunity for a comparison of tissue-specific changes in gene expression for donors spanning a range of age groups. Because multiple tissue types can be sourced back to the individual donor, our study allows for the investigation into how different tissues age within an individual, and how generalizable this multi-tissue aging profile may be in the study population. This study aims to develop a more comprehensive tissue-specific model to estimate biological age.