bioq: a unified, agent-native command-line interface to a fleet of AI drug-discovery methods

bioq is a unified command-line interface designed to streamline the use of various AI drug-discovery methods. Traditionally, each tool in the drug-discovery pipeline - ranging from structure prediction to ADMET analysis - comes with its own incompatible GPU software stack. This makes it challenging for researchers to chain multiple tools together seamlessly, as they often require conflicting software environments and access to expensive datacenter-class hardware that most labs lack.

bioq addresses this challenge by providing a dependency-light command-line client backed by bioq-services. bioq-services acts as an AC control plane gateway, along with a growing fleet of over 38 containerized drug-discovery tools. These tools span seven stages of discovery and six molecular modalities. The bioq CLI is self-describing and consistent across the entire fleet of computational tools, allowing researchers and coding agents to access any tool from a laptop with ease.

There's no need for local model code, CUDA setup, or cloud configuration. The bioq interface, gateway, and services architecture create an execution substrate for automated, agent-driven discovery. bioq and bioq-services are both open source, released under the MIT License and available on GitHub.

To use bioq, Python 3.10 is required with httpx as the only runtime dependency. Services run as Linux containers and can be self-hosted using provided local deployment scripts. Additionally, scripts for Alibaba Cloud Function Compute are available. Comprehensive install, quickstart, test data, and self-hosting instructions are included in the GitHub repositories. The bioq project is actively maintained and will continue to receive updates to features and services.