README.md

In gerbeldo/tidycell: CellID Data Analysis Inside the Tidyverse

rcell2

Functions to analyze Cell-ID's single-cell cytometry data in R, using a tidy framework.

rcell2's functionality is split into four packages:

• The main rcell2 package offers functions to load Cell-ID's output to data.frames, and image manipulation based on EBImage. A development version of this package is available in the rcell.dev branch.
• Cell-ID, the image segmentation software, has been wrapped in the rcell.cellid package. It offers functions to run CellID from R, and an rmarkdown template showcasing advanced functionality.
• The cell tiling and graphic filtering apps, built on R-Shiny and magick, are available in the rcell.magick package.
• The rcell2.examples package contains notebooks on general usage, and on several classification and analysis methods.

This package is very well tested in baker's yeast data, and R version 4+.

Preview

Provide a defocused brightfield image to CellID, and voila:

Segmentation of yeast cells in a single position.

The image is segmented, cells are identified and tracked over time, and features are computed from morphology and fluorescent signal distribution.

Time series images of one cell, showing different acquisition channels.

With Rcell2, you can load an analize the CellID results freely, using standard R packages.

Background corrected fluorescent signal concentration VS time, plotted with ggplot2.

Installation

Dependencies

Most of the dependencies are listed in the DESCRIPTION file, and should install automatically.

We suggest installing the tidyverse meta-package (and use it, you'll not regret it) and devtools:

install.packages(c("tidyverse", "devtools"))

In addition, install EBImage package (required to look at cells) by copying and running the following script:

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("EBImage")

Installing the package

Install using devtools, directly from github repo:

devtools::install_github("darksideoftheshmoo/rcell2")

devtools::install_github("darksideoftheshmoo/rcell2", ref = "rcell2.dev")  # Development version

The strongly recommended additional packages can be installed by following the instructions below.

Functions to install and run Cell-ID from R

These tools are available in the rcell.cellid package.

devtools::install_github("darksideoftheshmoo/rcell2", ref = "rcell2.cellid")  # Cell-ID

R-Shiny and magick tools for cytometry data

These tools are available in the rcell.magick package.

devtools::install_github("darksideoftheshmoo/rcell2", ref = "rcell2.magick")  # Shiny apps and Magick iamges

New features

Hu Moment functions for raw cell segmentation data

We implemented the Hu moments descriptors in R, and use them on masks generated by CellID. Note that the masks must be generated by the the CellID mask_mod branch either by TSV output or by encoding CellIDs in the pixel intensities of boundary and/or interior points.

We recommend using the rcell.cellid package to generate the required input.

K-means filtering functions

The kmeans algotrithm helps filter cells based on clustering of CellID's variables computed from morphological and fluorescence information.

Use k-means and check out images of cells in each cluster. Then, filter them easily by cluster number.

See: ?rcell2::kmeans_clustering

Todo

• ~~Plot of a 2D grid of "representative" single cell images in a scatterplot (similar to EBImage).~~ Implemented in cellSpread, cellSpreadPlot, and "Pics" type plot in shinyCell.
• ~~Generic shiny function for filtering points in a custom ggplot~~. Implemented in plotApp.
• 1D / histogram filtering support in the filtering app.
• Per-facet filtering in shinyCell.
• cellStrips support in shinyCell image viewer.

gerbeldo/tidycell documentation built on Aug. 15, 2022, 2:35 p.m.