bio-workflows-cytometry-pipeline

name	bio-workflows-cytometry-pipeline
description	End-to-end flow cytometry workflow from FCS files to differential analysis. Orchestrates compensation, transformation, gating/clustering, and statistical testing with CATALYST/diffcyt. Use when processing flow or mass cytometry data end-to-end.
tool_type	r
primary_tool	CATALYST
workflow	true
depends_on	["flow-cytometry/fcs-handling","flow-cytometry/compensation-transformation","flow-cytometry/gating-analysis","flow-cytometry/clustering-phenotyping","flow-cytometry/differential-analysis","flow-cytometry/doublet-detection","flow-cytometry/bead-normalization","flow-cytometry/cytometry-qc"]

Flow Cytometry Pipeline

Pipeline Overview

FCS Files ──> Compensation ──> Transformation ──> Gated/Clustered Data
                                                          │
                                                          ▼
                  ┌─────────────────────────────────────────────────┐
                  │            cytometry-pipeline                   │
                  ├─────────────────────────────────────────────────┤
                  │  1. Load FCS Files                              │
                  │  2. Compensation & Transformation               │
                  │  3. QC & Filtering                              │
                  │  4. Clustering (FlowSOM) or Gating              │
                  │  5. Dimensionality Reduction (UMAP)             │
                  │  6. Differential Abundance/State Analysis       │
                  │  7. Visualization                               │
                  └─────────────────────────────────────────────────┘
                                                          │
                                                          ▼
                      Differential Cell Populations + Markers

Complete R Workflow (CATALYST)

library(CATALYST)
library(diffcyt)
library(SingleCellExperiment)
library(flowCore)
library(ggplot2)

# === 1. SETUP PANEL AND METADATA ===
# Panel definition
panel <- data.frame(
    fcs_colname = c('FSC-A', 'SSC-A', 'CD45', 'CD3', 'CD4', 'CD8', 'CD19',
                    'CD14', 'CD56', 'HLA-DR', 'Ki67', 'IFNg'),
    antigen = c('FSC', 'SSC', 'CD45', 'CD3', 'CD4', 'CD8', 
                    
    marker_class        
                         



md  data.frame
    file_name  list.files pattern  
    sample_id  paste0 
    condition    each  
    patient_id  paste0  


cat nrowmd 


fcs_files  file.path mdfile_name
fs  read.flowSetfcs_files


fs_comp  compensatefs spilloverfs


sce  prepDatafs_comp panel md
                transform  
                cofactor    
                FACS  

cat ncolsce 



tablescesample_id


plotExprssce color_by  
ggsave width   height  


plotMDSsce color_by  
ggsave width   height  


cat
sce  clustersce
               features    
               xdim   ydim  
               maxK  
               seed  


tablecluster_idssce 


cat
sce  runDRsce dr   features  


plotDRsce dr   color_by  
ggsave width   height  

plotDRsce dr   color_by  
ggsave width   height  



plotExprHeatmapsce features   k  
                by   scale   bars  
ggsave width   height  


cluster_annotations  
      
      
      
      
      
    

scecell_type  cluster_annotationscluster_idssce 


cat


design  createDesignMatrixeisce cols_design  


contrast  createContrast   


res_DA  testDA_edgeRsce design contrast cluster_id  

da_results  as.data.framerowDatares_DA
da_results  da_resultsorderda_resultsp_adj 
cat
printda_results    


res_DS  testDS_limmasce design contrast
                        cluster_id  
                        markers_include  rownamesscerowDatascemarker_class  

ds_results  as.data.framerowDatares_DS
cat
sig_ds  ds_resultsds_resultsp_adj   
printsig_ds    



plotDiffHeatmapsce res_DA    fdr  
ggsave width   height  


plotAbundancessce k   by   group_by  
ggsave width   height  


da_resultssignificant  da_resultsp_adj  
ggplotda_results aesx  logFC y  log10p_adj color  significant 
    geom_pointsize   
    geom_hlineyintercept  log10 linetype   
    scale_color_manualvalues    
    theme_bw 
    labstitle  
ggsave width   height  


write.csvda_results  row.names  
write.csvds_results  row.names  
saveRDSsce 

cat
cat da_resultsp_adj   

flowCore + Manual Gating Workflow

library(flowCore)
library(flowWorkspace)
library(ggcyto)

# Load data
fs <- read.flowSet(list.files('data/', pattern = '\\.fcs$', full.names = TRUE))

# Compensation
comp_matrix <- spillover(fs[[1]])[[1]]
fs_comp <- compensate(fs, comp_matrix)

# Transformation
trans <- estimateLogicle(fs_comp[[1]], colnames(comp_matrix))
fs_trans <- transform(fs_comp, trans)

# Create GatingSet
gs <- GatingSet(fs_trans)

# Apply gates
gs_add_gating_methodgs alias  
                     pop   parent  
                     dims  
                     gating_method  
                     gating_args  K   target   

gs_add_gating_methodgs alias  
                     pop   parent  
                     dims  
                     gating_method  


autoplotgs 


gated_data  gs_pop_get_datags 

Python Alternative (FlowCytometryTools)

import flowkit as fk
import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler
from sklearn.cluster import KMeans

# Load FCS files
sample = fk.Sample('sample.fcs')

# Get data as DataFrame
data = sample.as_dataframe(source='raw')

# Compensation (if needed)
comp_matrix = sample.metadata['spill']
data_comp = np.dot(data, np.linalg.inv(comp_matrix))

# Arcsinh transformation
cofactor = 150  # For flow cytometry
data_trans = np.arcsinh(data_comp / cofactor)

# Clustering
scaler = StandardScaler()
data_scaled = scaler.fit_transform(data_trans)

kmeans = KMeans(n_clusters=10, random_state=42)
clusters = kmeans.fit_predict(data_scaled)

QC Checkpoints

Stage	Check	Action if Failed
Loading	All FCS files read	Check file integrity
Compensation	Spillover values reasonable	Recalculate
Transformation	Distributions normalized	Adjust cofactor
Events	>10K cells per sample	Check acquisition
Clustering	10-30 populations	Adjust K/resolution
DA	>3 replicates per group	Need more samples

Workflow Variants

CyTOF Data

# CyTOF-specific settings
sce <- prepData(fs, panel, md,
                transform = TRUE,
                cofactor = 5,  # CyTOF uses cofactor 5
                FACS = FALSE)  # Not flow cytometry

# Bead normalization should be done upstream (Fluidigm software)

Paired Design

# For paired samples (e.g., pre/post treatment)
design <- createDesignMatrix(ei(sce), cols_design = c('condition', 'patient_id'))

# Include patient as blocking factor
formula <- createFormula(ei(sce), cols_fixed = 'condition', cols_random = 'patient_id')
res_DA <- testDA_voom(sce, formula, contrast)

Related Skills

• flow-cytometry/fcs-handling - FCS file operations
• flow-cytometry/compensation-transformation - Data preprocessing
• flow-cytometry/gating-analysis - Manual gating
• flow-cytometry/clustering-phenotyping - Unsupervised clustering
• flow-cytometry/differential-analysis - Statistical testing
• flow-cytometry/doublet-detection - Remove doublet events
• flow-cytometry/bead-normalization - CyTOF EQ bead normalization
• flow-cytometry/cytometry-qc - Comprehensive QC
• single-cell/clustering - Related clustering methods