Fe·Rx

A nonlinear mixed-effects engine for pharmacometrics, written in Rust.

ferx fits models defined in comprehensible syntax, using state-of-art algorithms powered by a Rust backend. In any data science environment.

MIT — license

:::

How ferx is built

ferx-core is the Rust engine — it parses .ferx model files, runs the estimators, and produces fits. All numerical work happens here. It ships with a standalone command-line tool (the ferx CLI) that reads a .ferx file plus a CSV dataset and writes results to disk, so you can use the engine directly with no R involved.

ferx-r is a thin R wrapper around ferx-core. It exposes the same engine through R functions, hands data.frames to the core, and returns fit objects you can pipe into R’s plotting and reporting ecosystem.

Because ferx-core is a stable library, you (or anyone else) can build a different wrapper on top of it — Python, Julia, a web service — and get the exact same numerical results as the CLI and ferx-r.

Learn ferx

Get started
Install ferx and run the warfarin quickstart in under five minutes.

ferx-r
Introduction to FeRx and guide for working with FeRx from R.

ferx-core
Guiden and reference for working with FeRx using the CLI or Rust library.

Reference
Function-level documentation for every exported function in the ferx R package.