BioGAIP: A Scalable, User-Friendly and Robust LLM-Powered Multi-Agent System for Automated Multi-omics Tasks

We have released an updated version of the software. Please select "v1.3.1" from the zenodo navigation bar on the right to download it.

- **bioag-launcher-win32-x64-1.3.0.zip**: The executable file of BioLauncher for the Windows platform.
- **Quick Config.mp4**: How to quickly configure the running environment of BioLauncher, recommended steps (but optional)*.
- **BioGAIP-source-1.3.0.zip**: Source code of BioGAIP, including BioAG, BioWorker and BioLauncher.

- Empty LLM configuration file template: you can down LLM API configuretion template from here. This is a text file, you can edit it with notepad, vscode or other edit tools.

Alternatively, configuration can be performed graphically within the BioGAIP main interface, however, utilizing a configuration file is strongly recommended to ensure better reproducibility.

*The optimal running environment of the BioGAIP client on the Windows platform is to work together with WSL2. The Quick Config.mp4 provides guidance on how to configure it, this is recommended but optional. We have provided a bypass mode, allowing the software to run independently of WSL2. To do so, please skip  Quick Config.mp4, simply launch the `BioLancher` executable and select Bypass mode (running directly on Windows) in the very first step (please be aware that we provide only limited support for this mode).

Bypass mode requires the Microsoft Visual C++ 2015 Redistributable. While this is already installed on most Windows machines, you may need to install it if it's missing. 

You can find the installers in the root directory of the extracted archive:
* Run `VC_redist.x64.exe` (recommended for most 64-bit systems).
* Run `VC_redist.x86.exe` (for 32-bit systems).
(Tip: If you are unsure whether you already have it, it is perfectly safe to just run the installer anyway.)

For the mainstream Unix-like operating systems, the default version of glib should already meet the requirements if you run from source.

Setting up Bypass mode for the first time takes **approximately 20 minutes**. This is because the software needs to build and configure a local Python runtime environment in the background. Please be patient and do not close the program. This lengthy process **only occurs during the initial launch**.

We have built a wiki for BioGAIP, please visit: https://notebook.biogaip.top

Wiki password can be found in our manuscript

After the article is published, this code will also be open sourced at: https://github.com/zhangjy859/BioGAIP

Quick Start

Please refer to Quick_Start.docx

For macOS Users: We are actively working on providing standalone executables for macOS as well. However, due to a lack of testing platforms, this may take some time. Currently, macOS users will need to run the application from source.

Quick Evaluation Guide and Resources

To facilitate the evaluation of the software by peer reviewers and reduce the cost of evaluation, we provide quick evaluation resources established in our computing facilities. These resources include:

1. A series of pre-configured LLM APIs, including Qwen-max and Grok-fast-4.
2. A toy BioWorker environment containing a pre-defined public RNA-seq instance data.

To access these quick resources, please refer to the **Quick_Evaluation_Resource.zip** document, which will be archived as part of this Zenodo repository. Please note that this toy environment will be destroyed after the peer review process ends.  

The evaluation resources will be provisioned and uploaded to the Zenodo repository within three days after the manuscript enters the peer-review stage for cybersecurity reasons.

Change log

260403: Fixed the anomaly in the code that was causing issues with matching the parameters of the configuration file.

260408: Fixed bypass mode issues.

260417: Added a notify for the startup interface when the network is slow.

20250523: 

We have released an updated version of the software. Please select "v1.3.1" from the zenodo navigation bar on the right to download it.