Abstract
An important task of aging research is to find genes that regulate lifespan. However, identification of genes related to longevity (referred to as longevity genes hereafter) through web-lab experiments such as genetic screens is a tedious and labor-intensive activity. Developing an algorithm to predict longevity genes should facilitate aging research. In this paper, we systematically analyzed properties of longevity genes in Caenorhabditis elegans and found that, when compared to genes not yet known to be involved in longevity, known longevity genes display the following features: (i) longer genomic sequences and protein sequences, (ii) a stronger tendency to co-express with other genes during a transition from dauer state (an extremely long lifespan) to non-dauer state (a normal lifespan), (iii) significant enrichment in certain functions and RNAi phenotypes, (iv) higher sequence conservation, and (v) higher in several network topological features such as degrees in a functional interaction network. Based on these features, we developed an algorithm to predict longevity genes in C. elegans and obtained 243 novel longevity genes with a precision rate of 0.85. Some of the predicted genes have been validated by published articles or wet lab experiments. The contribution of each feature to the predicted results was also evaluated.

/Presenter : Hyun-Hee Hong

/Date : 2018.05.11