Data I/O API Reference

The data_io package provides tools for loading, preprocessing, and managing retinal datasets. It includes base classes for creating custom dataloaders and implementations for specific published datasets.

Overview

The data_io module is organized into:

• Base Data Classes — Core data containers (MoviesTrainTestSplit, ResponsesTrainTestSplit)
• Base Dataloader — MovieDataSet, MovieSampler, and dataloader factories
• Artificial Stimuli — Synthetic stimulus generation (chirp, moving bar)
• Cyclers — Multi-dataloader cycling utilities

Dataset Implementations

Each published dataset has its own subpackage with stimuli loading, response loading, and constants:

• Hoefling et al. 2024 — Mouse retinal ganglion cell responses to natural stimuli. From A chromatic feature detector in the retina signals visual context changes (eLife).
• Karamanlis et al. 2024 — Mouse and marmoset retinal ganglion cell responses to natural stimuli. From Nonlinear receptive fields evoke redundant retinal coding of natural scenes (Nature).
• Goldin et al. 2022 — Mouse retinal ganglion cell responses under varying visual contexts. From Context-dependent selectivity to natural images in the retina (Nature Communications).
• Maheswaranathan et al. 2023 — Primate and mouse retinal ganglion cell responses to natural scenes. From Interpreting the retinal neural code for natural scenes: From computations to neurons (Neuron).
• Sridhar et al. 2025 — Marmoset retinal ganglion cell responses to naturalistic movies and white noise. From Modeling spatial contrast sensitivity in responses of primate retinal ganglion cells to natural movies (bioRxiv).