ASAP: a web-based platform for the analysis and interactive visualization of single-cell RNA-seq data

Abstract

Motivation: Single-cell RNA-sequencing (scRNA-seq) allows whole transcriptome profiling of thousands of individual cells, enabling the molecular exploration of tissues at the cellular level. Such analytical capacity is of great interest to many research groups in the world, yet these groups often lack the expertise to handle complex scRNA-seq datasets.

Results: We developed a fully integrated, web-based platform aimed at the complete analysis of scRNA-seq data post genome alignment: from the parsing, filtering and normalization of the input count data files, to the visual representation of the data, identification of cell clusters, differentially expressed genes (including cluster-specific marker genes), and functional gene set enrichment. This Automated Single-cell Analysis Pipeline (ASAP) combines a wide range of commonly used algorithms with sophisticated visualization tools. Compared with existing scRNA-seq analysis platforms, researchers (including those lacking computational expertise) are able to interact with the data in a straightforward fashion and in real time. Furthermore, given the overlap between scRNA-seq and bulk RNA-seq analysis workflows, ASAP should conceptually be broadly applicable to any RNA-seq dataset. As a validation, we demonstrate how we can use ASAP to simply reproduce the results from a single-cell study of 91 mouse cells involving five distinct cell types.

Availability and implementation: The tool is freely available at asap.epfl.ch and R/Python scripts are available at github.com/DeplanckeLab/ASAP.

Contact: bart.deplancke@epfl.ch.

Supplementary information: Supplementary data are available at Bioinformatics online.

Fig. 1

ASAP pipeline. The figure depicts the complete pipeline, including tools, that is implemented in ASAP. The user starts by uploading a count matrix (or a normalized matrix) of gene expression after which either the default pipeline or different filtering algorithms can be selected. After the normalization step, the user can apply different dimensionality reduction methods to visualize the data in 2D or 3D. The user can interactively select samples, or run clustering algorithms to perform differential gene expression analysis. Finally, the selected gene list can be analyzed for enrichment in biological modules or pathways such as the Gene Ontology or KEGG. All tools are referenced in Supplementary Table S1