In tanaylab/MCView: A Shiny App for Metacell Analysis

knitr::opts_chunk$set(
    collapse = TRUE,
    comment = "#>"
)

Setup

First, let's load the MCView package:

library(MCView)

For this tutorial, we'll use a PBMC dataset that was pre-processed using the metacells python package.

dir.create("raw")
download.file("http://www.wisdom.weizmann.ac.il/~atanay/metac_data/PBMC_processed.tar.gz", "raw/PBMC_processed.tar.gz")
untar("raw/PBMC_processed.tar.gz", exdir = "raw")

This gives us two files: "raw/metacells.h5ad" (the metacell data) and "raw/cluster-colors.csv" (color assignments for cell types).

Import Dataset

The most common way to start with MCView is to directly import your dataset:

import_dataset(
    project = "PBMC", # This will create the project if it doesn't exist
    dataset = "PBMC",
    anndata_file = "raw/metacells.h5ad",
    cell_type_field = "type"
)

This single command: 1. Creates a project directory structure if it doesn't exist 2. Pre-processes the metacell dataset for viewing in the app 3. Adds an app.R file, making the project deployment-ready

The anndata_file points to the h5ad file we downloaded, and cell_type_field specifies the field in the h5ad object that contains pre-computed cluster assignments.

Note: Import may take a few minutes depending on the number of metacells. If it's too slow, set calc_gg_cor = FALSE to skip calculating gene correlations. This saves significant import time but makes this feature unavailable in the app.

Tip: Run compute_for_mcview in the metacells python package before importing to enable all MCView features.

MCView also supports importing datasets from the old R metacell package with import_dataset_metacell1.

Run the app

Once import is complete, you can run the app:

run_app(project = "PBMC", launch.browser = TRUE)

This opens a browser window with the app. You can customize the port, host, and browser launch:

run_app(project = "PBMC", port = 5555, host = "127.0.0.1", launch.browser = FALSE)

If you're already in the project directory, simply run:

run_app(launch.browser = TRUE)

Update annotations

After working with the initial metacell model, you might want to update the annotations:

1. Export your changes by pressing the "export" button in the upper left of the "Annotate" screen
2. Update the annotations with:

update_metacell_types("PBMC", "PBMC", "/path/to/metacell_types_file")

Where "/path/to/metacell_types_file" is the path to your exported file.

To only update cell type colors:

update_cell_type_colors("PBMC", "PBMC", "/path/to/cell_type_colors_file")

Working with metadata

You can add metadata for each metacell during import:

import_dataset(
    project = "PBMC",
    dataset = "PBMC",
    anndata_file = "raw/metacells.h5ad",
    metadata = "raw/metadata.csv"
)

The metadata should be a data frame (or file) with a column named metacell and additional annotation fields, which can be numeric or categorical.

Use metadata_colors to customize colors for each field.

To update metadata after import:

update_metadata(
    project = "PBMC",
    dataset = "PBMC",
    metadata = "new_metadata.csv"
)

update_metadata_colors(
    project = "PBMC",
    dataset = "PBMC",
    metadata_colors = new_metadata_colors
)

You can generate metacell metadata from cell metadata using cell_metadata_to_metacell or cell_metadata_to_metacell_from_h5ad.

Deploy your app

MCView projects are deployment-ready by default. The app.R file added during import means you can deploy your app by simply copying the entire project directory to a Shiny server.

Direct Deployment

Deploy to shinyapps.io:

rsconnect::deployApp(appDir = "PBMC")

Or deploy to any Shiny server by copying the "PBMC" directory to the server.

Advanced Deployment (Optional)

For additional customization (such as including MCView code in your bundle):

create_bundle(
    project = "PBMC",
    path = getwd(),
    name = "PBMC_bundle",
    self_contained = TRUE # Includes MCView code in the bundle
)

# Then deploy
rsconnect::deployApp(appDir = file.path(getwd(), "PBMC_bundle"))

Note: MCView keeps the metacell matrix in memory and requires around 1GB of RAM for small datasets like PBMC, and up to 2-4GB for larger datasets. Adjust your hosting service settings accordingly.

Advanced: Manual project creation

If you want to customize your project configuration before importing data, you can manually create a project first:

create_project("PBMC", title = "PBMC")

This opens a text editor with PBMC/config/config.yaml:

title: PBMC
tabs: ["QC", "Manifold", "Genes", "Markers", "Gene modules", "Diff. Expression", "Cell types","Annotate", "About"]
# selected_gene1: Foxc1
# selected_gene2: Twist1
selected_mc1: 1
selected_mc2: 2

After editing, proceed with import_dataset as shown earlier.

See the architecture vignette for a complete description of configuration parameters.

Docker

See the docker vignette for instructions on using the Docker image.

tanaylab/MCView documentation built on June 1, 2025, 8:08 p.m.