Skip to content

Latest commit

 

History

History
172 lines (140 loc) · 8.02 KB

index.md

File metadata and controls

172 lines (140 loc) · 8.02 KB
Raw
<style> p.caption { font-size: 0.8em; } </style>

scCustomize

CRAN Version CRAN Downloads license R-CMD-check issues DOI

scCustomize is an R package with collection of functions created and/or curated to aid in the visualization and analysis of single-cell data using R.

Installing scCustomize

Please see Installation page for full installation instructions.

Vignettes/Tutorials

See Vignettes for detailed tutorials of all aspects of scCustomize functionality.

Goals/About scCustomize

The goals of scCustomize are to:

1. Customize visualizations for aid in ease of use and create more aesthetic visuals.
2. Improve speed/reproducibility of common tasks/pieces of code in scRNA-seq analysis with a single or group of functions.

scCustomize aims to achieve these goals through:

  • Customized versions of many commonly used plotting functions (and some custom ones).
    To create greater flexibility in visualization and more aesthetic visuals by:
    • Altering default parameters for more intuitive plots (or at least I believe more intuitive). For instance: FeaturePlot(..., order = TRUE).
    • Wrapping commonly used ggplot2 post-plot themeing into function call. No more copy/paste of the same theme elements for every plot over and over (e.g., plot + scale_color_continuous(...) + ggtitle(...) + theme(plot.title = element_text(...), legend.position = ...) + guides(...)
    • Creating new plotting functions either: 1. as wrapper around Seurat function with parameters already specified (e.g., QC_Plot_Genes()) or 2. create new plots (e.g., Seq_QC_Plot_Reads_per_Cell() or Plot_Median_Genes()) or 3. both (e.g., QC_Plot_UMIvsGene(..., combination = TRUE)).
    • Adding additional parameters to existing plots inside new function (e.g., high and low cutoff parameters in QC_Plot_UMIvsGene())
  • Easy iterative plotting functionality.
    Many plotting functions can be easily automated with loops, apply, purrr etc. However, these can be intimidating to novice user and often can be made easier through wrapping into a function.
    • scCustomize contains a number of iterative plotting functions which contain extra parameters to specify file type, path, name and then render progress bar in console to track progress.
    • Returns either single PDF document or multiple plots of any valid file type (e.g., png, tiff, jpeg, pdf, etc).
  • Helper functions easily import multiple raw data types
    Import data functions are aimed at streamlining importing multiple files/samples with single function and/or importing files with “non-standard” file names.
    • Iterate import functions to simplify the import process across groups of files/samples.
    • Parallelize functions where possible to allow for dramatic speed improvements when import large number of samples simultaneously.
    • Provide easy wrapper functions to import files with output formats (e.g., CellBender) not supported by Seurat or other common R package.
  • Helper functions to simplify analysis with addition of new default parameters or wrapping multiple lines of code into single function.
    Goal is to both speed up and simplify coding and reduce the use of copy/paste of the same lines of code which is more likely to lead to errors in code reproducibility.

    • Example of adding new parameters: Adding the percentage of counts aligning to mitochondrial (and/or ribosomal) genes is common early step in analysis. scCustomize provides Add_Mito_Ribo() to simplify this. Basic use requires only one line of code and two parameters.

      Add_Mito_Ribo(object = obj_name, species = "Human")
      • Function already knows the defaults for Human, Mouse, Rat, Zebrafish, Drosophila, Marmoset, and Rhesus Macaque (submit a PR if you would like more species added!).

    • Example of wrapping many lines to one: Extracting the top 10 (or 15, 20, 25, etc) genes per identity after running Seurat::FindAllMarkers() is very common and scCustomize provides Extract_Top_Markers() function to simplify process.
      Using scCustomize function:

      markers_df <- FindAllMarkers(object = obj_name)

# Get vector/string
top10_list <- Extract_Top_Markers(marker_dataframe = markers_df)

# or for data.frame
top10_df <- Extract_Top_Markers(marker_dataframe = markers_df, dataframe = TRUE)

      Instead of tidyverse/base R:

      markers_df <- FindAllMarkers(object = obj_name)

# Get vector/string
top10_list <- markers_df %>%
  rownames_to_column("rownames") %>%
  group_by(cluster) %>%
  slice_max(n = 10, order_by = avg_log2FC) %>%
  pull("gene")

# or for data.frame
top10_df <- markers_df %>%
  rownames_to_column("rownames") %>%
  group_by(cluster) %>%
  slice_max(n = 10, order_by = avg_log2FC) %>%
  column_to_rownames("rownames")
  • Provide more informative error messages for many common issues
    Base R error messages resulting from error deep inside Seurat (or other package) function can sometimes be difficult to interpret, especially for users new to R.
    • scCustomize provides checks/warnings, using the cli/rlang packages, wrapped inside its functions to help and provide more informative error/warning messages. Two examples include:
    • Add_Mito_Ribo() will warn you if no mitochondrial or ribosomal features are found and won’t create new metadata column.
    • Rename_Clusters() will check and make sure the right number of unique new names are provided and provide one of two error messages if not before attempting to rename the object idents.

Support for Other scRNA-seq Object Formats (LIGER, SCE, etc)

Currently the package is primarily centered around interactivity with Seurat Objects with some functionality with LIGER objects and support for CellBender outputs.
If users are interested in adapting functions (or creating separate functions) to provide comparable functionality with SCE or other object formats I would be happy to add them. See below for more info on PRs.

Bug Reports/New Features

If you run into any issues or bugs please submit a GitHub issue with details of the issue.

  • If possible please include a reproducible example (suggest using SeuratData package pbmc dataset for lightweight examples.)

Any requests for new features or enhancements can also be submitted as GitHub issues.

  • Even if you don’t know how to implement/incorporate with current package go ahead a submit!

Pull Requests are welcome for bug fixes, new features, or enhancements.

  • Please set PR to merge with “develop” branch and provide description of what the PR contains (referencing existing issue(s) if appropriate).