Stardust: structural representations for refinement and beyond

Stardust is a library that defines Structure object that is fully featured for structural biology applications.

NOTE: This repo is a WIP, and should be considered hot lava code at the moment. Don't build on it. We are still working to get the concepts right!

Structure and Vision

Stardust implements two primary abstractions:

1. A structure module that helps manage topology, coordinate, B-factor, and occupancy information. It defines a powerful Structure object, as well as code that is crucial for converting and interoperating with different coordinate representations in use.

2. losslab. This specifies an interface for loss/likelihood functions that facilitates interoperability between different refinement projects. Because these loss functions depend on attributes of a Structure (for example, the Cartesian coordinate positions or B-factors), designing these interfaces for compatability with Structure enables maximum interoperability. Further, in addition to a Structure interface, valid losses will have an overloaded pure-array interface, which maximizes compatability.

Out of scope

Stardust does not generate or sample structures/coordinates.

Stardust assumes your are working with a discrete list of cartesian coordinates that represent atomic positions. Models that use densities, continous distributions, etc. are out of scope.