REGALS

REGALS deconvolves small-angle X-ray scattering (SAXS) data from evolving mixtures of biological macromolecules to extract component scattering profiles and concentration profiles when components and their temporal evolution are unknown.

Key Features:

• Regularized alternating least squares: Uses a regularized alternating least squares algorithm to perform matrix factorization of SAXS mixture data.
• Parameterization and regularization: Incorporates prior knowledge via parameterization and regularization constraints, including smoothness of profiles and maximum dimensions of species.
• Experimental versatility: Applied to SAXS datasets from chromatography-coupled SAXS (including anion-exchange coupled SAXS), ligand titrations, time-resolved mixing studies, and temperature-jump experiments.
• Implementation and numerical libraries: Implemented in MATLAB and Python with the Python implementation using numpy and scipy and supporting data import/visualization via h5py and matplotlib.

Scientific Applications:

• Deconvolution of evolving mixtures: Recover individual component scattering and concentration profiles from mixtures where neither components nor their evolution are known.
• Chromatography-coupled SAXS analysis: Separate overlapping species in chromatography-coupled SAXS experiments, including anion-exchange coupled SAXS.
• Ligand titration studies: Resolve component changes and binding-related scattering signals across ligand concentration series.
• Time-resolved experiments: Analyze time-resolved mixing and temperature-jump SAXS data to characterize dynamic structural changes.
• Structural biology of dynamic systems: Support investigation of interactions and structural transitions in biological macromolecules within complex mixtures.

Methodology:

REGALS applies a regularized alternating least squares algorithm with parameterization and regularization that incorporate prior knowledge such as profile smoothness and species maximum dimensions.