aind-behavior-services

aind-behavior-services

A repository containing code for data acquisition and processing for AIND behavior rigs.

---

Installation

The python package can be installed from pypi using the following command:

pip install aind-behavior-services

However, to use all the tasks and hardware that this package supports, you should over over the prerequisites and deployment sections.

Prerequisites

These should only need to be installed once on a fresh new system, and are not required if simply refreshing the install or deploying to a new folder.

• Windows 10 or 11
• Run ./scripts/install_dependencies.ps1 to automatically install dependencies
• The following dependencies should be manually installed:

  • Spinnaker SDK 1.29.0.5 (device drivers for FLIR cameras)

    • On FLIR website: Download > archive > 1.29.0.5 > SpinnakerSDK_FULL_1.29.0.5_x64.exe

---

Deployment

Install the prerequisites mentioned below. From the root of the repository, run ./scripts/deploy.ps1 to bootstrap both python and bonsai environments.

---

Generating valid JSON input files

One of the core principles of this repository is the strict adherence to json-schemas. We use Pydantic as a way to write and compile our schemas, but also to generate valid JSON input files. These files can be used by Bonsai (powered by Bonsai.SGen code generation tool) or to simply record metadata. Examples of how to interact with the library can be found in the ./examples folder.

---

Regenerating schemas

Once a Pydantic model is updated, updates to all downstream dependencies must be made to ensure that the ground-truth data schemas (and all dependent interoperability tools) are also updated. This can be achieved by running the regenerate command from the root of the repository. This script will regenerate all json-schemas along with C# code (./scr/Extensions) used by the Bonsai environment.

---

Contributors

Contributions to this repository are welcome! However, please ensure that your code adheres to the recommended DevOps practices below:

Linting

We use ruff as our primary linting tool.

Testing

Attempt to add tests when new features are added. To run the currently available tests, run python -m unittest from the root of the repository.

Lock files

We use uv to manage our lock files.

Versioning

Where possible, adhere to Semantic Versioning.

---

Project dependency tree

classDiagram
    class aind_behavior_curriculum {
        +Task
        +Curriculum
    }

    class aind_behavior_services {
        +Task (Subclasses)
        +Rig (maintains hardware library)
        +Session
        +Calibration (maintains device/calibration library)
        +Deployment instructions
        +Ecosystem documentation
    }

    class aind_behavior_experiment_launcher {
        +Launch experiment
        +Interfaces with external applications (e.g. Bonsai)
        +Interfaces with aind-services
    }

    class aind_behavior_some_task {
        +Concrete implementation of a task
        +Rig (Subclasses for some task)
        +Session
        +Task Logic (Subclasses for some task)
        +Maintains a task data-schema
        +Saves data in standard format
    }

    class aind_behavior_some_task_analysis {
        +Analysis code for some task
    }

    class aind_behavior_core_analysis {
        +Data ingestion
        +Data contract definition
        +Core analysis primitives
        +QC
    }

    aind_behavior_curriculum --|> aind_behavior_services : Subclasses Task
    aind_behavior_services --|> aind_behavior_some_task 
    aind_behavior_some_task --|> aind_behavior_some_task_analysis : Analysis
    aind_behavior_core_analysis --|> aind_behavior_some_task_analysis : Imports core analysis methods
    aind_behavior_some_task_analysis --|> aind_behavior_curriculum : Metrics[Task]
    
    aind_behavior_experiment_launcher --|> aind_behavior_some_task : Launches