TBA24 Analysis

Analysis pipeline for Tissue-Based Assay (TBA) immunofluorescence data examining autoantibody binding patterns in brain tissue.

Overview

This repository contains R scripts for analyzing and visualizing tissue-based assay data comparing autoantibody binding across different brain regions in cases and controls. The pipeline includes data preprocessing, principal component analysis (PCA), correlation analysis, and visualization.

Project Structure

TBA24_analysis/
├── scripts/           # R analysis scripts
│   └── archive/      # Older/exploratory scripts
├── data/             # Data files (not tracked in git)
├── plots/            # Output plots (not tracked in git)
└── README.md

Analysis Pipeline

1. Data Cleaning and Preparation

Script: data_cleaning.R

• Imports QuPath measurement data
• Extracts metadata from image filenames
• Aggregates fluorescence intensities by brain region
• Creates exploratory visualizations (bar plots, spider plots)

2. PCA Analysis

Scripts: PCA_TBA_function.R, cytokine PCA.r

• Flexible PCA function for different feature combinations
• Separate analyses for Cy5 (IgG) and FITC channels
• Options to include/exclude experimental controls
• Multiple PCAs on clinical, biomarker, and combined features

3. Correlation Analysis

Scripts: cytokine PANSS correlation.R, correlationsgraph.R

• Merges PCA results with clinical datasets
• Calculates correlations between PCA axes and clinical variables
• Generates correlation heatmaps

4. IgG Level Analysis

Script: igg_levels.R

• Analyzes serum and CSF IgG levels
• Calculates CSF/serum ratios
• Identifies and visualizes outliers

5. Hemoglobin Contamination Analysis

Script: hb samples.r

• Correlates hemoglobin (Hb) optical density with PCA scores
• Tests for blood contamination effects in CSF samples
• Compares Hb levels between cases and controls using parametric and non-parametric tests

6. Visualization

Scripts: clean plot.r, clean_heatmap.R, PCA_loadings.R

• Publication-ready PCA plots with custom colors
• Heatmaps of fluorescence intensities by brain region
• PCA loading analysis and interpretation

Requirements

R Packages

# Data manipulation
library(dplyr)
library(tidyr)
library(stringr)

# Visualization
library(ggplot2)
library(pheatmap)
library(viridis)
library(RColorBrewer)
library(ggrepel)
library(gridExtra)

# Analysis
library(reshape2)
library(tibble)

Usage

1. Set up data paths: Update the file path in data_cleaning.R to point to your local data directory
2. Run data cleaning: Execute data_cleaning.R to process raw QuPath data
3. Perform PCA: Run PCA_TBA_function.R for comprehensive PCA analysis
4. Analyze correlations: Execute correlation scripts to link PCA results with clinical variables
5. Generate visualizations: Run visualization scripts to create publication-ready figures

Data Format

Input Files

• QuPath measurements: CSV file containing ROI measurements with columns for DAPI, FITC, and Cy5 channels
• Clinical data: Dataset with subject demographics and biomarker measurements

Image Naming Convention

Images should follow this naming pattern: TBAxx_yy_PROJECT_ID

• xx: Experiment number
• yy: Slide number
• PROJECT: Project code (e.g., OHC, PMS)
• ID: Subject identifier

Output

• PCA plots showing group separation
• Heatmaps of regional fluorescence patterns
• Correlation matrices
• Bar plots with individual data points
• Spider/radar plots for regional patterns

Notes

• Outlier detection is performed on PCA results
• Group classifications: case, control, exp_control
• Both mean and standard deviation features can be included in analyses

Author

Analysis by Vasishta

License

This project is for research purposes.