Skip to content

README.md

Latest commit

 

History

History
308 lines (209 loc) · 5.52 KB

README.md

File metadata and controls

308 lines (209 loc) · 5.52 KB

TriNetX Neuropsychiatric Outcomes Analysis Pipeline

R pipeline for parsing TriNetX cohort exports, generating publication-ready figures, computing prospective statistical power, and exporting consolidated results tables for manuscript preparation.

Overview

This repository contains:

  • Raw TriNetX export files
  • An R analysis pipeline
  • Publication-ready figures
  • Consolidated results tables
  • Statistical power analyses

The workflow parses TriNetX MOA and Kaplan–Meier exports across three propensity-score-matched concussion cohorts and generates manuscript figures and supplementary analytic outputs automatically.

Repository Structure

.
├── data/
│   ├── Concuss_1x_5_year_V12_424528/
│   ├── Concuss_2x_5_year_V12_(1)_(1)_677193/
│   └── Concuss_3x_5_year_V12_327389/
│
├── R/
│   └── make_figures.R
│
├── results/
│   ├── figure2_forest.*
│   ├── figure3_power.*
│   ├── power_table.csv
│   ├── results.csv
│   ├── figure_captions.txt
│   └── session_info.txt
│
└── README.md

Requirements

R Version

  • R 4.3 or newer recommended

Required Packages

The script automatically loads:

ggplot2
dplyr
readr
stringr
purrr
scales
grid
gtable
patchwork

Install missing packages with:

install.packages(c(
  "ggplot2",
  "dplyr",
  "readr",
  "stringr",
  "purrr",
  "scales",
  "gtable",
  "patchwork"
))

Running the Pipeline

From the repository root directory, run:

Rscript R/make_figures.R

The script will automatically:

  • Parse all TriNetX outcome tables
  • Build consolidated results datasets
  • Apply BH-FDR correction
  • Compute prospective statistical power
  • Generate publication-ready figures
  • Export all outputs to ./results

No additional configuration is required if the repository structure is preserved.

Input Data

The pipeline expects three comparison directories inside ./data:

Comparison Description
Concuss_1x_5_year_V12_424528 Single concussion vs orthopedic fracture controls
Concuss_2x_5_year_V12_(1)_(1)_677193 Two concussions vs one concussion
Concuss_3x_5_year_V12_327389 Three concussions vs two concussions

Each directory must contain TriNetX exports including:

  • MOA tables
  • KM tables
  • associated graph CSVs
  • baseline characteristic files

The analysis script specifically parses:

Outcome_*_Result_a_MOA_table.csv
Outcome_*_Result_b_KM_table.csv

for outcomes 1–8.

Outcomes Included

Internalizing / Neurodevelopmental

  • Mood disorders
  • Anxiety disorders
  • ADHD
  • Unspecified psychosis

Substance Use

  • Generalized substance use disorder
  • Alcohol-related disorders
  • Cannabis-related disorders
  • Opioid-related disorders

Statistical Methods

Relative Risk Analysis

The script extracts:

  • Relative risks (RR)
  • Hazard ratios (HR)
  • 95% confidence intervals
  • Log-rank p-values

from TriNetX exports.

Multiple Testing Correction

Benjamini–Hochberg false discovery rate correction is applied across all 24 hypothesis tests.

Prospective Power Analysis

Prospective statistical power is computed using:

  • two-sided α = 0.05
  • minimum clinically important RR = 1.20
  • normal approximation to the two-proportion z-test

using matched cohort sample sizes and observed reference-group event rates.

Output Files

All outputs are written to:

./results/

Main Tables

File Description
results.csv Consolidated analytic dataset
power_table.csv Prospective power calculations

Figures

Figure 2 — Forest Plot

Exports:

figure2_forest.svg
figure2_forest.pdf
figure2_forest.png
figure2_forest.tiff

Displays:

  • Relative risks
  • 95% confidence intervals
  • BH-FDR significance indicators

Figure 3 — Prospective Power Analysis

Exports:

figure3_power.svg
figure3_power.pdf
figure3_power.png
figure3_power.tiff

Displays:

  • Statistical power across outcomes and comparisons
  • Conventional 80% adequacy threshold

Additional Outputs

File Description
figure_captions.txt Manuscript-ready figure captions
session_info.txt Full R session metadata

Console Output

The script prints:

  • Figure captions
  • Comparisons with power < 0.80
  • Comparisons with power < 0.50

Example:

Cells with power < 0.80:
- CX3_vs_CX2 | Opioid-related | power=0.412

Figure Specifications

Figures are exported at:

  • Width: 260 mm
  • Height: 120 mm
  • Resolution: 300 DPI

Formats include:

  • SVG
  • PDF
  • PNG
  • TIFF (LZW compression)

Reproducibility

The analysis includes:

  • deterministic random seed (set.seed(20260421))
  • session metadata export
  • standardized plotting themes
  • fully scripted data processing

Notes

  • The script assumes consistent TriNetX export formatting.
  • Missing or malformed files will stop execution with informative errors.
  • Helvetica is used as the default figure font family.

License

Add your preferred license here.

Example:

MIT License

Citation

If used in academic work, please cite the associated manuscript or repository.