A catalogue of putative cis-regulatory interactions between long non-coding RNAs and proximal coding genes based on correlative analysis across diverse human tumors

Abstract

Antisense transcripts and other long non-coding RNAs are pervasive in mammalian cells, and some of these molecules have been proposed to regulate proximal protein-coding genes in cis. For example, non-coding transcription can contribute to inactivation of tumor suppressor genes in cancer, and antisense transcripts have been implicated in the epigenetic inactivation of imprinted genes. However, our knowledge is still limited and more such regulatory interactions likely await discovery. Here, we make use of available gene expression data from a large compendium of human tumors to generate hypotheses regarding non-coding-to-coding cis-regulatory relationships with emphasis on negative associations, as these are less likely to arise for reasons other than cis-regulation. We document a large number of possible regulatory interactions, including 193 coding/non-coding pairs that show expression patterns compatible with negative cis-regulation. Importantly, by this approach we capture several known cases, and many of the involved coding genes have known roles in cancer. Our study provides a large catalog of putative non-coding/coding cis-regulatory pairs that may serve as a basis for further experimental validation and characterization.