reach is a small Python library for creating forelimb reaching tasks for
mice or other rodents used for neuroscientific research. It has been developed
for my own implementation of a Raspberry Pi-based visually-guided reaching task
using one or two reaching targets and visual stimuli -- both head-fixed and
freely-moving -- but has been designed to work with arbitrary 'backends' that
may implement other stimulus or reach target configurations. reach is
concerned with controlling hardware, responding to events during the task, and
organising the resulting data in a format that makes analysis easy.
The code and data representation follows a hierarchical object-oriented structure:
| Object | Role |
Cohort |
Manages one or more Mouse instances.
Useful for data handling and plotting. |
Mouse |
Manages one or more Session instances.
Useful for analysing single animals and running training sessions. |
Session |
Manages a training session's behavioural data and operates the training sessions. |
Data is saved per animal in a JSON file. This data can be loaded as follows:
from reach import Mouse
data_dir = '/path/to/your/data/'
mouse = Mouse.init_from_file(
mouse_id='Mouse1',
data_dir=data_dir,
)A training session can be run, appending its results to the data stored in
Mouse. You would then want to save the data back to JSON to keep the new
data:
mouse.train(backend)
mouse.save_data_to_file(data_dir)backend here is what interfaces with the hardware controlling the task.
Details on creating custom backends is here.