Fast Mode

CASSIA's Fast Mode offers a streamlined, one-line solution for running the complete analysis pipeline. This mode combines annotation, scoring, and annotation boost for correcting low quality annotations in a single function call, using optimized default parameters.

Basic Usage

runCASSIA_pipeline(
    output_file_name = "my_analysis",
    tissue = "brain",
    species = "human",
    marker = marker_data,
    max_workers = 4
)
R

Add CASSIA results back to the seurat object

seurat_corrected <- add_cassia_to_seurat(
  seurat_obj = seurat_corrected, # The seurat object you want to add the CASSIA results to
  cassia_results_path = "/FastAnalysisResults_scored.csv", #where the scored results saved, specify the path
  cluster_col = "celltype", # Column in Seurat object with cell types
  cassia_cluster_col="True Cell Type" # Column in the scored results with the true cell types
)

# This will add six new columns to the seurat object:the genearl celltype, all three subcelltypes, the mostly likely celltype, the second likely celltype, the third likely celltype, and mixed celltype,and the quality score of each cell type.
R

Full Parameter Options

runCASSIA_pipeline(
    # Required parameters
    output_file_name,     # Base name for output files
    tissue,               # Tissue type (e.g., "brain", "blood")
    species,              # Species (e.g., "human", "mouse")
    marker,               # Marker file from findallmarker, path or the data obejct
    
    # Optional parameters with defaults
    max_workers = 4,      # Number of parallel workers
    
    # Model configurations
    annotation_model = "gpt-4o",             # Model for annotation
    annotation_provider = "openai",         # Provider for annotation
    score_model = "anthropic/claude-3.5-sonnet",  # Model for scoring, it is highly recommended to use a different and more powerful model for scoring
    score_provider = "openrouter",         # Provider for scoring
    annotationboost_model="anthropic/claude-3.5-sonnet", # Model for annotation boost
    annotationboost_provider="openrouter", # Provider for annotation boost
    
    # Analysis parameters
    score_threshold = 75,     # Minimum acceptable score
    additional_info = NULL    # Additional context information
)
R

Output Files

The pipeline generates a folder which contains the following files:

Annotation results csv files
Scored results csv files
Basic CASSIA report
Annotation boost report

Performance Tips

For optimal performance, adjust max_workers based on your system's CPU cores
Use additional_info to provide relevant experimental context
Monitor score_threshold to balance stringency with throughput

Next we introduce each function in detail...