FARNA: knowledgebase of inferred functions of non-coding RNA transcripts

Abstract

Non-coding RNA (ncRNA) genes play a major role in control of heterogeneous cellular behavior. Yet, their functions are largely uncharacterized. Current available databases lack in-depth information of ncRNA functions across spectrum of various cells/tissues. Here, we present FARNA, a knowledgebase of inferred functions of 10,289 human ncRNA transcripts (2,734 microRNA and 7,555 long ncRNA) in 119 tissues and 177 primary cells of human. Since transcription factors (TFs) and TF co-factors (TcoFs) are crucial components of regulatory machinery for activation of gene transcription, cellular processes and diseases in which TFs and TcoFs are involved suggest functions of the transcripts they regulate. In FARNA, functions of a transcript are inferred from TFs and TcoFs whose genes co-express with the transcript controlled by these TFs and TcoFs in a considered cell/tissue. Transcripts were annotated using statistically enriched GO terms, pathways and diseases across cells/tissues based on guilt-by-association principle. Expression profiles across cells/tissues based on Cap Analysis of Gene Expression (CAGE) are provided. FARNA, having the most comprehensive function annotation of considered ncRNAs across widest spectrum of human cells/tissues, has a potential to greatly contribute to our understanding of ncRNA roles and their regulatory mechanisms in human. FARNA can be accessed at: http://cbrc.kaust.edu.sa/farna.

Figure 1.

Brief description of the pipeline used to infer cell/tissue-specific functions of miRNA and lncRNA transcripts. First, FANTOM5 CAGE data for primary cells and tissues was collected. The miRNA, lncRNA, TF and TcoF transcripts that are expressed in different cells/tissues were used for further analysis in cell/tissue-specific manner. The TF binding site (TFBS) models from TRANSFAC and HOCOMOCO databases were used to predict TFBSs on promoter regions of miRNA and lncRNA transcripts. Only TFs and their associated TcoFs that are expressed in the considered cell/tissue together with the ncRNA transcript are used to infer statistically significant cell/tissue-specific functions of the transcript.

Figure 2.

Expression of of RNA transcripts in different tissues/cells. Subfigure (A) shows scatter plot with contour overlay and (B) shows the box plot for the fraction of transcripts from miRNA and lncRNA having expression >1 TPM in different tissues/cells. Subfigure (C) and (D) highlight the distribution of expression for miRNA and lncRNA transcripts in several tissues. Vertical red dashed line highlights the one TPM (normalized TPM using the relative log expression (RLE) method in edgeR) threshold.

Figure 3.

Basic modules of FARNA. There are four basic modules in FARNA. Data Sources: The repository for all TFs, TcoFs, CAGE expression data from different cells/tissues, and transcripts for both miRNA and lncRNA. CAGE data was used to filter out low-expressed ncRNA transcripts, TFs, TcoFs. FARNA Associations: This module considers the association of TFs with TcoFs to RNA transcripts and identifies statistically enriched functions related to RNA transcripts in a cell/tissue-specific manner. FARNA DB: This module indexes all associated function annotation using Elasticsearch platform. FARNA Web Interface: Web interface for users to explore the FARNA annotated function and expression profile of RNA transcripts in different cells/tissues.