Symphony Compare (Optional)

Overview

Note: Symphony Compare requires OpenRouter as the provider. This is the only supported provider for this module.

symphonyCompare is an advanced module that orchestrates multiple AI models to compare cell types with automatic consensus building. It conducts a comprehensive cell type comparison using multiple AI models in parallel, automatically triggering discussion rounds when models disagree on the best matching cell type. Think of it as a virtual panel of expert biologists debating and reaching consensus.

Quick Start

results <- symphonyCompare(
  tissue = "peripheral blood",
  celltypes = c("T cell", "B cell", "NK cell", "Monocyte"),
  marker_set = c("CD3", "CD4", "CD8", "CD19", "CD20", "CD16", "CD56", "CD14"),
  species = "human"
)

cat("Consensus:", results$consensus, "\n")
cat("Confidence:", sprintf("%.1f%%", results$confidence * 100), "\n")
R

Input

Marker genes: A character vector or comma-separated string of marker genes from CASSIA's previous results or your own analysis
Candidate cell types: A vector of 2-4 cell types to compare
Tissue context: The tissue type being analyzed
Species: The species of the sample (default: human)

Parameters

Required Parameters

Parameter	Type	Description
`tissue`	character	The tissue type being analyzed (e.g., "blood", "brain", "liver")
`celltypes`	character vector	A vector of 2-4 cell types to compare
`marker_set`	character vector/string	A list or string of marker genes

Optional Parameters

Parameter	Type	Default	Description
`species`	character	`"human"`	The species of the sample
`model_preset`	character	`"budget"`	Pre-configured model ensemble. `"premium"`: High-performance ensemble (Gemini 3 Pro, Claude Sonnet 4.5, GPT-5.1, Grok 4). `"budget"`: Cost-effective models (DeepSeek V3.2, Grok 4 Fast, Kimi K2, Gemini 2.5 Flash)
`enable_discussion`	logical	`TRUE`	If TRUE, models will "discuss" and reconsider their votes if initial consensus is not reached
`generate_report`	logical	`TRUE`	Whether to generate an HTML report of the analysis
`verbose`	logical	`TRUE`	Whether to print progress messages to the console

Advanced Parameters

Parameter	Type	Default	Description
`max_discussion_rounds`	integer	`2`	Maximum number of discussion rounds allowed
`consensus_threshold`	numeric	`0.8`	The confidence threshold required to declare consensus (0-1)

Output

The function returns a list containing consensus results and generates output files.

Return Value:

results: List of all model responses and scores
consensus: The consensus cell type (if reached)
confidence: Confidence level of the consensus (0-1)
csv_file: Path to the generated CSV file with detailed results
html_file: Path to the generated interactive HTML report (if enabled)
summary: Summary statistics of the comparison