OsteoSort 1.5.0

Computerized osteometric sorting application built with R/Shiny and Julia. OsteoSort uses statistical methods to compare skeletal measurements against reference populations, aiding in the reassociation of commingled remains.

Key Features:

• Pair-matching — statistical comparison of bilateral skeletal elements
• Articulation — assessment of joint congruence between adjacent bones
• Osteometric sorting by regression — size-based reassociation using OLS regression
• Interactive Plotly visualizations with CSV export
• PostgreSQL-backed reference populations (ARDS)

Architecture

Layer	Technology
Frontend	R/Shiny UI
Backend (statistical)	R + Julia (compiled shared library)
Database	PostgreSQL (ARDS)
Deployment	Docker (two-stage build)
Julia compilation	PackageCompiler.jl (create_library)

Prerequisites

• Docker
• A running PostgreSQL instance with the ARDS osteometry schema
• A .env file inside OsteoSort/ with database credentials:

  DB_HOST=<host>
  DB_PORT=<port>
  DB_USER=<user>
  DB_PASS=<password>
  DB_NAME=<database>
  

Installation

git clone https://github.com/dpaa-gov/OsteoSort
cd OsteoSort
docker build -t osteosort .
docker run --restart=on-failure:10 --name=osteosort -d -p 4001:3838 osteosort
docker network connect app_bridge osteosort 

The app will be available at http://localhost:4001/OsteoSort.

Local Development (Without Docker)

Requirements

• R 4.x with packages listed in Dependencies
• Julia 1.11+ (for building the shared library)
• GCC (for building the C shim)
• PostgreSQL client library (libpq-dev on Debian/Ubuntu)
• .env file in OsteoSort/ with DB credentials (see Prerequisites)

Build the Shared Library (one-time)

# Build libosj.so
julia --project=OSJ build/create_library.jl

# Build C shim
gcc -shared -fPIC -o build/r_osj_shim.so build/r_osj_shim.c \
    -L dist/libosj/lib -losj -Wl,-rpath,$(pwd)/dist/libosj/lib

Run

LD_LIBRARY_PATH=dist/libosj/lib:dist/libosj/lib/julia Rscript start_dev.R

The app will open at http://127.0.0.1:4001.

Project Structure

OsteoSort/
├── Dockerfile             # Two-stage: builder compiles .so, runtime is lean
├── start_dev.R            # Local dev server launcher
├── shiny-server.conf
├── OsteoSort/             # Shiny application
│   ├── server.r           # Server entry point (loads osj.r, calls osj_load())
│   ├── ui.r               # UI entry point
│   ├── R/                 # Analytical R functions
│   │   ├── osj.r          # Shared library interface (dyn.load + wrappers)
│   │   ├── ttest.r        # T-test analysis (calls osj_ttest)
│   │   └── reg.test.r     # Regression analysis (calls osj_regsl)
│   ├── server/            # Server modules (reference, single, files, etc.)
│   ├── ui/                # UI modules
│   ├── extdata/           # Config files (articulation_config, etc.)
│   └── www/               # Static assets (CSS, JS, images)
├── OSJ/                   # Julia analytical package
│   ├── Project.toml
│   └── src/
│       ├── OSJ.jl         # Module definition
│       └── c_api.jl       # @ccallable wrappers for R .C() interface
├── build/                 # Build scripts
│   ├── create_library.jl  # PackageCompiler library build
│   ├── library_precompile.jl
│   └── r_osj_shim.c      # C shim for init_julia ABI bridging
└── dist/                  # Build output (gitignored)
    └── libosj/
        └── lib/libosj.so

Dependencies

R

Package	Purpose
shiny	Web framework
htmltools	HTML generation
DT	Interactive data tables
dplyr	Data manipulation
shinyalert	Alert dialogs
DBI	Database interface
RPostgres	PostgreSQL driver
dotenv	Environment variable loading
plotly	Interactive plots

Julia (OSJ package)

Package	Purpose
Statistics	Statistical functions
Optim	Optimization
Rmath	R math distributions
GLM	Generalized linear models

Acknowledgments

• Alex Moore — UI styling suggestions and design inspiration

Citation

Lynch, J.J. 2026 OsteoSort. Computerized Osteometric Sorting. Version 1.5.0. Defense POW/MIA Accounting Agency, Offutt AFB, NE.

TODO

1. User beta testing
2. Replace .env file with injected environment variables

License

GNU General Public License v2.0