---

• Authors: Vijay Nagarajan PhD
• Affiliation: Laboratory of Immunology, NEI/NIH
• Contact: [email protected]

---

SCassist is an R package that utilizes a combination of statistical calculations and LLM-based insights to guide users through the complex process of single-cell RNA-seq data analysis. The package aims to provide recommendations, annotations, and interpretations, leading to efficient and insightful results.

Features:

• Recommendations for quality control and filtering
• Recommendations for the most appropriate Normalization method
• Analysis of variable features to identify processes driving the cell-to-cell variation
• Analysis of top PCs to glean insights in to processes driving the system
• Recommendations for optimal number of PCs to use for downstream analysis
• Recommendations for suitable range of resolution values for clustering
• Marker gene analysis and cell type prediction with detailed reasoning
• KEGG pathway and GO enrichment analysis and integration, providing deeper insights in to system understanding

Benefits:

• Automated Recommendations: Receive tailored recommendations for key parameters and analysis choices based on your specific data characteristics.
• LLM-Powered Insights: Leverage the power of LLMs to interpret complex data, uncover hidden patterns, and generate insightful summaries and explanations.
• Intuitive Interface: Easy-to-use functions with clear documentation and examples, making advanced analysis accessible to researchers of all levels.
• Confidential: Option to use a local LLM server to keep your data and analysis confidential.
• Cost effective: If taking advantage of google or openai models, use pay-as-you-go low cost API call options.

Installation:

# Install the necessary packages
install.packages("visNetwork")
install.packages("httr")
BiocManager::install("clusterProfiler")

# Install the devtools package if you don't have it
install.packages("devtools")

# Install SCassist from GitHub
devtools::install_github("NIH-NEI/SCassist")

LLM Server Setup:

• SCassist Local Ollama Server Setup
• Install Ollama following instructions here: https://github.com/ollama/ollama
• Start ollama desktop application or from command line (ollama serve)

# Install rollama package to use the local ollama llm server
install.packages("rollama")

# Download the model (in R)
pull_model("llama3")

• SCassist Remote Google Server Setup - obtain api-key following the instructions here: https://ai.google.dev/gemini-api/docs/api-key

• SCassist Remote OpenAI Server Setup - obtain api-key following the instructions here: https://platform.openai.com/api-keys

Example Usage:

Download example data: NK, CD4+ and CD8+ T cells from LCMV infected Ifng - CTCF binding site mutant mice - GSM6625298_scRNA_LCMV_Day4_CD4_CD8_NK_WT_filtered_feature_bc_matrix.h5

# Load the SCassist and Seurat packages
library(SCassist)
library(Seurat)

# Load the downloaded example file
KO <- Read10X_h5("GSM6625298_scRNA_LCMV_Day4_CD4_CD8_NK_WT_filtered_feature_bc_matrix.h5", use.names = T)

# Create seurat object
KO <- CreateSeuratObject(counts = KO[["Gene Expression"]], names.field = 2,names.delim = "\\-")

# Set api_key_file variable
api_key_file = "api_key_from_google.txt"

# Recommend quality control filters using Gemini (online)
qc_recommendations <- SCassist_analyze_quality("KO", llm_server="google", api_key_file = api_key_file)

# Recommend quality control filters using OpenAi GPT model (online)
qc_recommendations <- SCassist_analyze_quality("KO", llm_server="openai", api_key_file = api_key_file)

# Recommend quality control filters using Llama3 (local)
qc_recommendations <- SCassist_analyze_quality("KO", llm_server="ollama")

# ...and many more functions!

Tutorial Datasets:

• PBMCs from Birdshot Uveitis Patient
• PBMCs from Healthy Human Control
• NK, CD4+ and CD8+ T cells from LCMV infected WT mice
• NK, CD4+ and CD8+ T cells from LCMV infected Ifng - CTCF binding site mutant mice

Tutorials:

Step-by-step tutorials documenting the full workflow for the example datasets are provided below:

• Single Cell RNA-Seq Analysis of PBMC's from a healthy human Vs an Uveitis patient, using SCassist - an AI-Based Workflow Assistant

• Single Cell RNA-Seq Analysis of NK, CD4+ and CD8+ T cells isolated from LCMV infected WT and Ifng - CTCF binding site mutant mice, using SCassist - an AI-Based Workflow Assistant

Old Seurat Workflows, for comparison:

The below workflows are the original, standard workflow versions. We used these old versions to evaluate our new SCassist based workflow.

• Single Cell RNA-Seq Analysis of PBMC's from a healthy human Vs an Uveitis patient - Old workflow

• Single Cell RNA-Seq Analysis of NK, CD4+ and CD8+ T cells isolated from LCMV infected WT and Ifng - CTCF binding site mutant mice - Old workflow

Documentation:

Detailed documentation for each function, including parameters, usage, and expected outputs, is available through the ? help function in R. For example, run ?SCassist to known about all the included functions, run ?SCassist_analyze_quality to learn about the syntax, parameters, expected inputs, defaults and outputs about the function that analyzes the quality of your single cell data and recommends filtering options.

Supporting Files/Scripts:

• SCassist Template Prompts
• SCassist Semantic Similarity Evaluation Script

License:

The license for this package can be found in the LICENSE file within the package directory.