title	pbmc_score.md
author	Taylor Johnson
date	2026-10-1
output	github_document

PBMC Score

Background/Overview

This repository contains scripts for assessing consistency of peripheral blood mononuclear cell (PBMC), or other cell type, isolations between processing technicians and provides metrics for establishing baselines for future processing technicians. Parallel processing of a minimum of two technical replicates per research participant visit is required to effectively use this repository, and is visualized in an interactive Shiny app.

Whole blood (CSF, etc.) – and the cell populations derived from it – reflects the biological and technical environments through which it natively exists, is acquired, processed, and then analyzed. The goal of clinical research sample processing is to provide a frozen timepoint of biological information with the greatest biologically-derived signal to noise. Thus, management of biorepository quality over time is important as each procedure from acquisition to analysis has the potential to introduce unwanted confounders, with each technician influencing their own touch, every step in the process. Confounders have the potetntial to affect experiment, project, and study outcomes.

Given: (i) The innate variabilities within each participant (ii) The sensitivity of research technologies deployed to assess this variation (single cell RNA seq, timsTOF MS-GC, etc.) (iii) The unknown magnitude of effect within each unique cohort to be determined (iv) The exploratory nature of foundational studies leading to hypothesis-driven experiment designs The importance of consistency in sample processing cannot be overstated.

Statistics derived from simple raw cell counts per technitian (live, dead, total) are compared between phlepotomy tube replicates per isolation to create a "PBMC Score" aggregate which is then used to compare isolation quality between technitians. PBMC Score is composed of two major metric means: (i) Total cell viability of tube replicates (ii) Viability and total cell counts between tube replicates. These two values capture quality and consistency and are averaged with equal weight to obtain a final composite metric for comparison.

Mathematically, PBMC score is calculated as follows: mean(mean(viability_tube_A, viability_tube_B) + mean(inter_tube_viability, inter_tube_total))

The purpose is to (i) Establish baseline PBMC biorepository metrics (ii) Provide a reference to guague all future-processing technitians' quality by. This importantly informs the average quality of the total biorepository storage currently on hand, and creates a tool for managing quality of all future samples to be banked, promoting a more consistent sample pool amassed over time (through decades and technicians), future-proofing the system.

Project Aims

Aim 1: Generate baseline statistics: Assess between current processing technicians using reference_pbmc_score.Rmd

Aim 2: Generate new-user metrics: Assess single data-point values versus reference statistics using new_user_pbmc_score.Rmd

Analysis Plan

Analysis:

This software: reference_pbmc_score.Rmd, is used to work up values entered in your reference dataset (see example_reference_dataset.csv for example formating) to your generate baseline statistics representative of the biorepository total. PBMC Score will be calculated per processing technitian for comparison analysis, and the aggregated dataset (reference_pbmc_score_data.rds) will be used for subsequent analysis in new_user_pbmc_score.Rmd to compare single-datapoint values against previously-generated baseline statistics.

Comparisons between processing technicians are performed using a combination of parametric and non-parametric methods, depending on metric type (cell count, proportion). PBMC Score, mean viability, inter-tube viability, and inter-tube total cell count are compared using two-sided Wilcoxon rank-sum (Mann–Whitney U) tests to account for non-normality and limited sample sizes. Raw cell count metrics (total, live, and dead cell counts) are evaluated using parametric, mean-based comparisons, with all results visualized in a Shiny app.

Data Structure

Project directory:

project_directory
├── data
    └── example_reference_dataset.csv  
├── output
├── README.md
└── scripts 
    ├── new_user_pbmc_score.Rmd
    └── reference_pbmc_score.Rmd

Responsibilities and Acknowledgements

Taylor Johnson is responsible for all content related to this repository