MERFISH Processing

Overview

This page describes the data processing workflow for the MERFISH mouse colitis dataset. The raw data is obtained through the Bioconductor MerfishData package and undergoes filtering, quality control, and normalization to generate the fully-processed analysis-ready reference (healthy) and query (DSS-induced colitis) datasets.

Data Source

The raw dataset is available through MerfishData R/Bioconductor package:

Package: MerfishData

Dataset: MouseColonIbdCadinu2024() — Mouse colon tissue profiled with MERFISH technology

Citation: Cadinu et al. (2024). Charting the cellular biogeography in colitis reveals fibroblast trajectories and coordinated spatial remodeling. Cell, 187(8), 2010-28.

Genes: 943 genes profiled across healthy and DSS-treated samples

Processing Pipeline

The workflow is executed through a single R script:

File: R/merfish/Data_QC_and_Processing.R

Step 1: Data Loading and Sample Selection

Load the MERFISH dataset from MerfishData:

Code 
library(MerfishData)
spe <- MouseColonIbdCadinu2024()

Then filter for:

  • Slice: Tissue slice 2
  • Single mouse per condition:
    • Healthy (Day 0): Mouse ID 082421_D0_m6
    • DSS-treated (Day 9): Mouse ID 062221_D9_m3

Step 2: Quality Control Filtering

Apply adaptive QC thresholds to remove low-quality cells:

  • Library size: Remove cells with abnormally low total UMI counts
  • Feature detection: Remove cells detecting very few genes
  • Mitochondrial content: Remove cells with excessive mitochondrial gene expression

Step 3: Biological Filtering (Reference Only)

Reference dataset only: Remove any cells labeled as inflammation-associated from the healthy reference using the addMergedCellTypes() function including:

  • IAE (Inflamed Epithelial)
  • IAF (Inflamed Fibroblast)
  • IASMC (Inflamed SMC)

as those were considered artifacts from the manual-marker based annotation performed by the authors on the overall clustering of cells across timepoints.

Step 4: Normalization

Re-calculate log-normalized counts (logcounts) on the quality-filtered data using library size factors.

Step 5: Diagnostic-Aware PCA

Uses the addReferencePCA() function to:

  • Identify highly variable genes (HVGs) in both reference and query separately
  • Take their union to capture both stable biological signals and disease-specific signals
  • Run PCA on this diagnostic HVG set (50 components) for both datasets

This ensures comparability between the two conditions while preserving signals unique to inflammation.

Step 6: Cell Type Annotation Merging

Uses the addMergedCellTypes() function with MERFISH-specific mapping rules to collapse fine-grained tissue annotations (tier2) into broader, unified cell type categories (tier2_merged). Enables consistent interpretation across analyses.

Cell type groups include: Fibroblast, Inflamed Fibroblast, Smooth Muscle, Inflamed SMC, Epithelial, Inflamed Epithelial, Stem/TA, Immune lineages (Neutrophil, Macrophage, etc.), Endothelial, and others.

Step 7: Formatting & Cleaning

  • Subset metadata to essential columns (cell type, sample info)
  • Convert assays to in-memory sparse matrices (required for downstream tools like scVI)
  • Remove alternative experiments to reduce file size
  • Add standardized cell names

Fully-processed analysis-ready datasets

Both output objects are in SpatialExperiment format with:

  • Assays: counts (raw) and logcounts (log-normalized)

  • Spatial data: X/Y coordinates retained

  • Metadata: Cell-level annotations including:

    • Cell type (tier2 and tier2_merged)
    • Mouse ID and sample type

  • Reduced dimensions: PCA (50 components) computed on selected HVGs

Dimensions:

  • Reference (Healthy baseline, Day 0): 27,140 cells; 943 genes
  • Query (DSS-induced colitis, Day 9): 29,040 cells; 943 genes

Downloading Pre-processed Data

Rather than running this pipeline from scratch (slow), fully processed datasets can be obtained directly following the instructions on the Data retrieval page (fast).

Full Script

For complete implementation details, see:

  • Main script: R/merfish/Data_QC_and_Processing.R

Supporting functions:

  • R/auxiliary/addReferencePCA.R — Diagnostic-aware PCA using union of reference and query HVGs
  • R/auxiliary/addMergedCellTypes.R — Cell type categorization with MERFISH-specific mappings