Date of Award

10-6-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Information Science

First Advisor

Fusheng Tang

Abstract

Aging in eukaryotes is an extremely complex process associated with many different phenotypes. The aging process operates on many different scales or levels within an organism, from the molecular to a whole complex physiology. There are many different aging mechanisms, e.g. in yeast many different processes underlie chronological (measured in hours) lifespan and replicative lifespan (RLS) (measured in generations). To understand what causes aging and what occurs during aging, scientists have collected multiple sets of genome-wide data, such as transcriptomes, from the model organism, the budding yeast *Saccharomyces cerevisiae*. These data are especially valuable for the application of novel bioinformatic technique to better understand mechanisms of aging and predict lifespan under different conditions. To extract knowledge from genome-wide data, I utilized several supervised machine learning algorithms (including feature selection and feature importance ranking) as well as graph theory including guilt-by-association. Time series plots were used for visualizing gene expression levels and protein abundances. RLS was focused on rather than chronologically lifespan because there are more datasets available for RLS. Using these methods, I crafted predictive models for aging genes and RLS-extending interventions such as aging subclass predictors and lifespan predictors through classification and regression, respectively. This work focuses on three types of predictors. Subsequently, discoveries are made with the statistical and learning algorithms. The first model (a lifespan predictor) is trained on predicting the lifespan based on genotype, environment and combinations thereof. It is useful for predicting lifespan-extending interventions at the population level. The second model (an age predictor) is trained on predicting age given features measured on individuals. This is useful for identifying biomarkers of aging and to determine the effects of interventions on the level of individuals. The third model predicts functions/regulations of biological entities regarding the aging process, based on heterogeneous data such as ontologies and diverse omics, including time-series gene expression profiles (which can be visualized as plots), and Linked Data (i.e. structured data that be interlinked and becomes more useful through semantic queries). This third model is used to understand the role of genes and proteins as well as perhaps other entities such as small molecules including lipids and other metabolites. The functions of proteins, many of which are still unknown, especially those involved in yeast lipid metabolism and its regulation, can be predicted with this model. Among the predicted longevity genes, many of them are involved in lipid metabolism such as phosphatidylcholine biosynthesis, the Kennedy pathway, regulation of the transcription factor OverProducer of Inositol 1 (OPI1) which encodes Opi1 (a negative regulator of genes involved in lipid synthesis), Opi1-target genes, the mevalonate pathway, and sphingolipid metabolism. My predictions are validated by published wet-lab studies on lipid metabolisms, such as the role of the OxySterol binding protein Homolog OSH6, a oyxsterol-binding protein with multiple lipid ligands, and OPI1 regulation in aging [Tang et al., unpublished]. Therefore, my predictive models are applicable to studying aging in other organisms, including the human. The novel aspects of this research are for instance that 1) aging is investigated systematically in an unbiased data-driven approach, 2) lifespan is predicted as continuous values, 3) age is predicted by combining multiple omics data, and 4) functions and regulations of biological entities like genes are predicted with high confidence from heterogeneous data sources. Through research for this dissertation, I discovered that genetics is the most important feature of lifespan determination. Phenotypic features related to lipids and membranes, such as vacuolar morphology and autophagic activity are important for lifespan determination according to the best performing models. An age predictor based on transcriptomics and proteomics can highly accurately determine age. Its selected features are associated with both translation and lipid metabolism. Among the top selected features are transcripts of genes which, when deleted, exhibit abnormal vacuolar morphology as well as genes which are targets of Opi1. Opi1 itself and its regulators were found to be differentially regulated post-transcriptionally and post-translationally. Lastly, a function predictor for genes was created that achieved exceptional accuracy in classifying aging genes into aging-supressors and gerontogenes. It learned for instance that *piecemeal autophagy of the nucleus* is strongly predictive for aging-suppressor genes while *cytoplasmic translation* is strongly predictive for gerontogenes.

Download

Share

COinS