Irfan Alibay - Towards MDAnalysis 3.0 | SciPy 2024

Learn about MDAnalysis 3.0's upcoming features, performance improvements, and community initiatives for molecular dynamics simulation analysis in Python at SciPy 2024.

Key takeaways
  • MDAnalysis is a 17-year-old open-source Python library for analyzing molecular dynamics simulations, with support for most MD engines and file formats

  • Key features include:

    • Reading/writing multiple trajectory formats
    • Analysis tools and distance calculations
    • On-the-fly trajectory transformations
    • Extensible architecture for custom analyses
    • NumPy array-based data structures
  • Moving towards MDAnalysis 3.0 (2025) with focus on:

    • Performance improvements through Cython/C++ optimization
    • Better handling of large datasets (multi-TB)
    • Improved distance calculations via vectorization
    • Enhanced parallelization capabilities
    • Slimmer, more maintainable codebase
  • New “MDAnalysis Kits” initiative:

    • Similar to scikit model
    • Encourages community-developed downstream packages
    • Weekly testing of submitted kits
    • Central registry for discovery
  • Community growth efforts include:

    • Google Summer of Code mentorship
    • Workshops and outreach programs
    • User guides and documentation
    • 206+ contributors to date
    • Community events and meetings
  • Technical improvements:

    • New data structures at C-level
    • Enhanced interoperability with other libraries
    • Improved distance calculations library
    • Better periodic boundary condition handling
    • GPU support for larger systems