QSPQuest
QSPQuest is an open-source engine that simulates pharmacology models and estimates their parameters. A model declares its equations, units, parameters and dosing events, and the engine compiles and solves it. It is published under the Apache License 2.0 at unibio-intelligence/qspquest.
QSPQuest is the engine inside UbiBiologics modeling. A scenario run from the UbiBiologics model library is computed by QSPQuest compiled to WebAssembly, in your browser, and runs submitted to the workspace are computed on UniBio Intelligence servers by a service built on the same engine. The same engine can be installed and run on your own computer, from Python, R, the command line or a Model Context Protocol (MCP) client.
This page records what a public reader can reach and sends you to the repositories for the procedures, so that the two sets of instructions cannot drift apart.
Install
Prebuilt wheels are published on the Python Package Index (PyPI) for CPython 3.10 and newer, covering Linux (x86_64 and ARM64), macOS (Apple Silicon and Intel) and Windows (x86_64):
python3 -m pip install qspquest
A wheel needs no Rust toolchain or compiler. Building from source does: the published documentation states Rust 1.98.1 with Cargo and a native compiler. The installation guide covers the optional extras, the R interface and the source builds.
Use it from an MCP client
A container image of the QSPQuest MCP server is published on the GitHub Container Registry as
ghcr.io/unibio-intelligence/qspquest-mcp. It includes a synthetic model, so a dose
simulation can be run without installing Rust or downloading a model library. The server
listens on your own machine, and a client on another machine cannot reach it. The
agent and container quickstart
gives the exact commands, the stdio setup and the expected results.
The runtime, notebook and R images are not published. They are built from the public source when they are needed.
Tutorials
The engine repository holds tutorials for first runs, oral pharmacokinetic (PK) analysis, fitting published theophylline concentrations, paired-endpoint fitting, population estimation with log-scale intervals, visual predictive checks and MCP use:
Model library
unibio-intelligence/qspquest-models is the open repository of models for QSPQuest, published under the Apache License 2.0. Each model comes with its archive, runnable presets and recorded evidence. Its README states that the library contains 72 models, of which 62 simulate, 5 replay published results and 5 are for inspection only, together offering 644 ready-to-run experiments. The model catalog lists them by topic, and each model carries a model card that states its scope.
The UbiBiologics model library is a separate, hosted collection that runs on the same engine. The two have different contents and different counts; neither is a subset of the other.
Validation
Validation results for QSPQuest are published at unibio-intelligence/qspquest-validation. That repository is licensed differently from the engine and the model library: it is published under the GNU General Public License version 3.
Data use
Examples in the engine and library documentation use fixed or synthetic parameters that illustrate a method. The published documentation states the boundary directly: matching a reference does not make a model clinically suitable, and the examples are not for choosing clinical doses. Read a model's stated scope and units before using its results.