docs/3otherformat.md

In SGDDNB/ShinyCell: Shiny Interactive Web Apps for Single-Cell Data

title: | | Tutorial for other supported file formats (h5ad / loom / SCE) author: | | John F. Ouyang date: "Feb 2021" output: pdf_document: default html_document: toc: true toc_depth: 2 toc_float: collapsed: false fontsize: 12pt pagetitle: "3otherformat"

ShinyCell supports all four common single-cell data formats, namely h5ad, loom, SCE and Seurat. In our other tutorials, we focused on using Seurat objects as inputs. Here, we will explain how to use the other supported file formats to create shiny apps from single-cell data objects.

Overall, the process is identical to the other tutorials except that the file path is required for h5ad / loom files and the SCE object itself is supplied for SingleCellExperiment objects. To further change the aesthetics and ordering of metadata, please refer to the Tutorial for changing ShinyCell aesthetics and other settings. To include multiple datasets into one shiny app, please refer to the Tutorial for creating a ShinyCell app containing several single-cell datasets.

h5ad file as input

For using h5ad file as input, we will demonstrate using the scRNA-seq of endocrine development in the pancreas, taken from Bastidas-Ponce et al. Development (2018). The dataset is taken from the scVelo package tutorial.

library(ShinyCell)
getExampleData("h5ad")

inpFile = "endocrinogenesis_day15.h5ad"
scConf = createConfig(inpFile)
makeShinyApp(inpFile, scConf, 
             default.gene1 = "Pcsk2", default.gene2 = "Dcdc2a",
             default.multigene = c("Pcsk2", "Dcdc2a", "Ank", "Gng12", "Top2a",
                                   "Pak3", "Tmem163", "Nfib", "Pim2", "Smoc1"),
             shiny.dir = "shinyAppH5ad/", 
             shiny.title = "ShinyCell h5ad Example") 

Running the above code generates a shiny app in the shinyAppH5ad/ folder, looking like this:

loom file as input

The loom file format is a hdf5-based file format that is commonly used in cell atlases e.g. Human Cell Atlas. We would like to emphasize again that ShinyCell is a visualisation tool and not an analysis tool with the aim of providing sharable visualisation of already-analysed single-cell datasets. We assume that common single-cell analysis such as dimension reduction and clustering has been performed. Thus, loom files downloaded directly from cell atlas data portals may not have these analysis performed and ShinyCell will not work accordingly.

To demonstrate using a loom file as input, we will be using the scRNA-seq of hematopoietic differentiation, taken from Setty et al. Nature Biotechnology (2019). Dimension reduction and clustering analysis of this dataset has been performed using the ASAP pipeline; key: xr7ne3.

library(ShinyCell)
getExampleData("loom")

inpFile <- "xr7ne3_dim_reduction_13225_output.loom"
scConf = createConfig(inpFile, meta.to.include = 
                        c("_Depth","_Detected_Genes","_Mitochondrial_Content",
                          "_Protein_Coding_Content","_Ribosomal_Content",
                          "_clust_1_seurat","bundle_version","donor_organism.sex"))

# Make "_clust_1_seurat" categorical and default
scConf = makeMetaCategorical(scConf, "_clust_1_seurat", inpFile)
scConf[ID == "_clust_1_seurat"]$fRow = 4
scConf = modDefault(scConf, "_clust_1_seurat", "bundle_version")

makeShinyApp(inpFile, scConf, 
             shiny.dir = "shinyAppLoom/", 
             shiny.title = "ShinyCell loom Example") 

Note that loom files do not necessarily store the categories / levels information for categorical single-cell metadata. ShinyCell tries to circumvent this by automatically factoring all metadata containing text. However, there are cases where the single-cell metadata are integers and it is difficult to automatically detect if these metadata are meant to be treated in a discrete or continuous manner. In this loom file, the seurat clusters _clust_1_seurat is one such example. To make the seurat clusters a categorical metadata, we ran the function makeMetaCategorical(). We further modified the number of rows in the legends of the seurat clusters _clust_1_seurat metadata to 4 and also set the metadata to be plotted by default.

Running the above code generates a shiny app in the shinyAppLoom/ folder, looking like this:

SCE object as input

For using SCE object as input, we will demonstrate using the Seurat object used in the other tutorials. The Seurat object can be easily converted to a SCE object and the SCE object is processed by ShinyCell in the same way as its Seurat counterpart. Here, we reuse the code from the quick start guide.

library(Seurat)
library(ShinyCell)

getExampleData()                       # Download example dataset (~200 MB)
seu = readRDS("readySeu_rset.rds")
sce = as.SingleCellExperiment(seu)
scConf = createConfig(sce)
makeShinyApp(sce, scConf, gene.mapping = TRUE,
             shiny.title = "ShinyCell Quick Start") 

SGDDNB/ShinyCell documentation built on Jan. 25, 2024, 3:19 p.m.