Abstract

We are interested in understanding the codes that drive the functions of long noncoding RNAs (lncRNAs) in mammalian cells. Comparative analysis has played instrumental roles in allowing us to understand the rules underlying protein and microRNA function, but it has been difficult to apply the existing methods to study lncRNAs because of scarcity of lncRNA annotations and rapid lncRNA evolution. To build a catalog of lncRNAs in vertebrates, we recently collected RNA -seq and 3P-seq data and developed robust methods for lncRNA identification in 17 species (Hezroni et al., 2015). We also developed methods for identifying lncRNA homologs based on subtle sequence homology and conservation of neighboring genes and genomic context. We then focused on the evolutionary origins of those lncRNAs that are found only in mammals and traced ~5% of them to protein-coding genes that have lost their coding potential before the rise of mammals. These pseudogene-derived lncRNAs are associated with features that set them apart from other lncRNAs, such as broader expression domains. I will also describe our use experimental perturbations in human and mouse cells undergoing differentiation to dissect lncRNA functions and the sequence and structure elements supporting them.