ISG Simulator

A Shiny app for simulating and comparing statistical designs commonly used in safety pharmacology studies. The app helps users explore the operating characteristics (power, type I error, bias, coverage) of several trial designs under user-specified assumptions, and visualize how power changes with sample size and effect size.

🌐 Live app: https://bing-liu.shinyapps.io/safety-pharm-design-simulater/

Designs supported

Tab	Design
p1	Parallel
p2	Parallel with repeated measures (Parallel-RM)
p3	Crossover (XOV)
p4	Sequential (SEQ)
p5	Ascending dose (ASC)
p6	Modified ascending dose (Modified-ASC)

For each design, the app provides three sub-tabs:

1. Design layout — preview of one simulated trial (animals, periods, treatments, baseline, response).
2. Operating characteristics — runs n_sim simulations and reports power, type I error, bias, CI coverage from a gt table.
3. Power analysis — closed-form power and sample-size curves (using pwr_ana_hc_* helpers), with optional adjustment for baseline measurements.

Repository layout

ISG_simulator/
├── shiny_standalone/                 # The full standalone Shiny app (deployable)
│   ├── app.R                         # Entry point
│   ├── app_helper.R                  # Path helpers (app_root, app_source, app_file)
│   ├── ui/                           # ui_p0.R … ui_p6.R + pics/
│   ├── server/                       # server_p0.R … server_p6.R
│   ├── functions/                    # pwr_ana_hc_*.R + plot/ helpers
│   ├── script/                       # import_power_tb_sim.R + functions/
│   ├── outputs/                      # Pre-computed simulation results (.rds)
│   ├── deploy_app.R                  # rsconnect deployment driver
│   ├── publish_app.sh                # Wrapper around deploy_app.R
│   └── README.md                     # App-specific deployment notes
├── PR.sh                             # Pull-request workflow helper
└── ISG_simulator.Rproj               # RStudio project file

The app is fully self-contained: every source() resolves inside shiny_standalone/, so the folder can be copied or deployed on its own.

Running locally

From R / RStudio:

setwd("shiny_standalone")
shiny::runApp()

From a shell:

cd shiny_standalone
R -e 'shiny::runApp(launch.browser = FALSE)'

Required R packages

shiny, shinydashboard, mathjaxr, bslib, gt, dplyr, pwr, ggplot2, plotly, tidyverse, rstatix, emmeans, ggpubr, lme4, parallel, doParallel, rsconnect.

Deploying to shinyapps.io

A token from https://www.shinyapps.io/admin/#/tokens must be registered once per machine via rsconnect::setAccountInfo(). After that:

cd shiny_standalone
./publish_app.sh

publish_app.sh clears the cached deployment metadata and runs deploy_app.R, which reads RSCONNECT_ACCOUNT, SHINYAPPS_TOKEN, and SHINYAPPS_SECRET from the environment.

Contributing

This repository follows the pull-request workflow documented in PR.sh:

1. Create / switch to my-feature-branch tracking origin/main.
2. Pull, commit changes, push the feature branch.
3. Open a PR on GitHub and merge it there.
4. Switch back to main, pull, delete the local feature branch.