CellView

A ShinyApp to visualize and explore single cell datasets

A live version of this app is hosted here: https://mbolisetty.shinyapps.io/CellView/

A preprint describing this software is available on bioRxiv.

Introduction

CellView reads expression, dimesionality reduction/clustering, and feature annotation data objects from an .Rds file, and provides functionality to quickly explore and interactively analyze single cell transcriptomic data.

• View 3D representation of your dataset.
• Analyze (co-)expression patterns within and among specific clusters.
• Compute differential gene expression analysis on the-fly using interactive sample selection.

Usage

The primary structure of the .Rds file comprises three dataframes, with the following object names and column names. These structures will be updated in the future to be more flexible but currently the following naming conventions are required.

• log2cpm - Your (N x M) genes vs. cells expression matrix. Gene names need to be in Ensembl gene id format (e.g. ENSG, ENSM).
• tsne.data – Your dimensionality reduction and sample clustering information. This dataframe contains M rows and 4 columns: V1, V2, V3, dbCluster.
  • V1, V2, V3 store the 3 dimensional representation of your data, e.g. from t-SNE, PCA, etc.
  • dbCluster contains numerical cluster assignments. The row names of this data frame must correspond to the column names of your expression matrix.
• featuredata - A dataframe representing gene annotations with row names in ensembl gene id format. The following 2 columns are required:
  • Chromosome.Name - integers representing chromosome numbers.
  • Associated.Gene.Name - Gene symbol Other columns are allowed but not utilized. The number of rows can be larger than the number of genes (N) in your expression matrix, but beware of duplicate gene names with unique ENSGIDs. You can use one of the provided files under CellView/FeatureData or create your own.

Sample code to generate an .Rds file for upload to CellView

options(stringsAsFactors = FALSE, row.names = 1, as.is = T)
log2cpm <- read.csv('Data/Expression.csv', check.names = F)
featuredata <- read.csv('Databases/HG19_v74_FeatureData.csv', sep = ',')
tsne.data <- read.csv('Data/TNSE_dbscan.csv')

save(log2cpm, featuredata, tsne.data, file = 'Filename.Rds')