Changelog

Changelog#

This document tracks all notable changes to scDataset.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]#

[0.2.0] - 2025-08-13#

Added#

  • Core scDataset class with flexible sampling strategies

  • Sampling strategies:

    • Streaming - Sequential sampling without shuffling

    • BlockShuffling - Block-based shuffling for locality

    • BlockWeightedSampling - Weighted sampling with blocks

    • ClassBalancedSampling - Automatic class balancing

  • Support for multiple data formats:

    • NumPy arrays

    • AnnData objects

    • HuggingFace Datasets

    • PyTorch Datasets

    • Any object with __getitem__ and __len__

  • Performance optimizations:

    • Block-based data fetching

    • Configurable fetch factors

    • Multiprocessing support

  • Customization features:

    • Custom fetch callbacks

    • Custom batch callbacks

    • Fetch and batch transforms

  • Comprehensive documentation and examples

  • Test suite with >90% coverage

  • GitHub Actions CI/CD pipeline

Technical Details#

  • Minimum Python version: 3.8

  • Core dependencies: torch >= 1.2.0, numpy >= 1.17.0

  • Compatible with PyTorch DataLoader

  • Supports both eager and lazy data loading patterns

Known Issues#

  • Performance may vary depending on storage backend (SSD vs HDD)

Note

This changelog will be updated with each release. See the GitHub releases page for the most up-to-date information.

Contents