IRIS-EDA: An integrated RNA-Seq interpretation system for gene expression data analysis

Abstract

Next-Generation Sequencing has made available substantial amounts of large-scale Omics data, providing unprecedented opportunities to understand complex biological systems. Specifically, the value of RNA-Sequencing (RNA-Seq) data has been confirmed in inferring how gene regulatory systems will respond under various conditions (bulk data) or cell types (single-cell data). RNA-Seq can generate genome-scale gene expression profiles that can be further analyzed using correlation analysis, co-expression analysis, clustering, differential gene expression (DGE), among many other studies. While these analyses can provide invaluable information related to gene expression, integration and interpretation of the results can prove challenging. Here we present a tool called IRIS-EDA, which is a Shiny web server for expression data analysis. It provides a straightforward and user-friendly platform for performing numerous computational analyses on user-provided RNA-Seq or Single-cell RNA-Seq (scRNA-Seq) data. Specifically, three commonly used R packages (edgeR, DESeq2, and limma) are implemented in the DGE analysis with seven unique experimental design functionalities, including a user-specified design matrix option. Seven discovery-driven methods and tools (correlation analysis, heatmap, clustering, biclustering, Principal Component Analysis (PCA), Multidimensional Scaling (MDS), and t-distributed Stochastic Neighbor Embedding (t-SNE)) are provided for gene expression exploration which is useful for designing experimental hypotheses and determining key factors for comprehensive DGE analysis. Furthermore, this platform integrates seven visualization tools in a highly interactive manner, for improved interpretation of the analyses. It is noteworthy that, for the first time, IRIS-EDA provides a framework to expedite submission of data and results to NCBI's Gene Expression Omnibus following the FAIR (Findable, Accessible, Interoperable and Reusable) Data Principles. IRIS-EDA is freely available at http://bmbl.sdstate.edu/IRIS/.

Fig 1. IRIS-EDA integrated functions.

Comparison of IRIS-EDA and six other DGE analyses and visualization tools regarding available features, integrated tools, visualizations, and analyses.

Fig 2. Correlation analyses.

(A) Interactive correlation heatmap generated from single-cell gene expression data; (B) Scatterplot generated by selected a cell in the interactive correlation heatmap; (C) Sample distance matrix showing Euclidean distances between samples, along with hierarchical clustering.

Fig 3. Principal component analysis, multidimensional scaling, and t-distributed Stochastic Neighbor Embedding.

(A) PCA plot showing the first two principal components; (B) MDS plot showing the first two MDS coordinates; and (C) t-SNE plot showing the first two t-SNE coordinates.

Fig 4. Clustering and biclustering.

(A) Sample dendrogram and color bar representing optimized identified clusters for the WGCNA method of clustering on the scRNA-Seq example data. The dendrogram shows the 2- and 4-cell samples clustering together, with the Late Blastocysts forming a unique cluster. (B) The first three biclusters were generated using QUBIC on the IRIS-EDA server. The first two biclusters (69 and 49 genes) show the grouping of Oocyte, Zygote, and 2- and 4-cell samples and Oocytes, Zygote, and 2-, 4-, and 8-cell samples, respectively. The third bicluster (52 genes) separates the Late Blastocysts from the other samples. These three biclusters demonstrate the expression similarity between the Oocyte, Zygote, and multi-cell samples relative to the Late Blastocyst samples over numerous gene sets.

Fig 5. DGE overview.

An overview of the number of DEGs determined using DESeq2 on the IRIS-EDA server. Cell-type Seven is compared against the other six cell types based on the number of up- and down-regulated genes. The Seven and Four comparison shows the highest number of DEGs of all comparisons, followed by Seven and Five comparison. The other four comparisons show similar numbers of DEGs, with all comparisons showing at least as many down-regulated genes as up.

Fig 6. DGE overview.

(A) MA plot for the Seven and Four cluster comparison with particular genes highlighted in the results table and corresponding location in the figure; (B) Volcano plot for the Seven and Four cluster comparison with particular genes highlighted in the results table and corresponding location in the figure; (C) The searchable, interactive table corresponding to both the MA plot and Volcano plot, showing results of the DGE analysis from the user-selected DGE tool.