ExpressionDB: An open source platform for distributing genome-scale datasets

Abstract

RNA-sequencing (RNA-seq) and microarrays are methods for measuring gene expression across the entire transcriptome. Recent advances have made these techniques practical and affordable for essentially any laboratory with experience in molecular biology. A variety of computational methods have been developed to decrease the amount of bioinformatics expertise necessary to analyze these data. Nevertheless, many barriers persist which discourage new labs from using functional genomics approaches. Since high-quality gene expression studies have enduring value as resources to the entire research community, it is of particular importance that small labs have the capacity to share their analyzed datasets with the research community. Here we introduce ExpressionDB, an open source platform for visualizing RNA-seq and microarray data accommodating virtually any number of different samples. ExpressionDB is based on Shiny, a customizable web application which allows data sharing locally and online with customizable code written in R. ExpressionDB allows intuitive searches based on gene symbols, descriptions, or gene ontology terms, and it includes tools for dynamically filtering results based on expression level, fold change, and false-discovery rates. Built-in visualization tools include heatmaps, volcano plots, and principal component analysis, ensuring streamlined and consistent visualization to all users. All of the scripts for building an ExpressionDB with user-supplied data are freely available on GitHub, and the Creative Commons license allows fully open customization by end-users. We estimate that a demo database can be created in under one hour with minimal programming experience, and that a new database with user-supplied expression data can be completed and online in less than one day.

Fig 1. The ExpressionDB user interface is designed to showcase RNA-seq data with straightforward visuals.

Dot plots represent expression levels of different transcripts in different samples, with error bars representing +/- S.E.M. Results can be filtered by gene symbol, gene name/description, or Gene Ontology (GO) terms.

Fig 2. ExpressionDB supports a number of advanced options to filter data.

(A) By ticking the “Advanced Filtering” option, the user may choose to examine ranges of expression levels, as well as choose the reference sample for calculating fold change between any two samples. ExpressionDB allows filtering based on q-values, allowing the user to browse through statistically significant features. (B) Additional visualization methods, including downloadable tables, heatmaps, and volcano plots can also be accessed here.

Fig 3. ExpressionDB supplies four built-in visualization methods.

Example (A) Volcano Plots, (B) Gene Comparisons, (C) Heatmaps, and (D) Principal Component Analysis are shown here.