ShiftCrypt

ShiftCrypt encodes per-residue NMR chemical shift data into a single biophysical index for residue-level analysis and biophysical alignment of proteins.

Key Features:

• Neural network auto-encoder: Leverages an auto-encoder architecture to compress per-amino-acid NMR chemical shifts into a single value.
• Residue-level biophysical index: Produces a ShiftCrypt value per amino acid residue that represents its biophysical state and is independent of amino acid type.
• Chemical shift representation: Retains multi-dimensional and cross-correlated information from NMR chemical shifts that reflect residue dynamics and conformation.
• Biophysically-based pairwise alignment: Aligns proteins using Dynamic Time Warping (DTW) on ShiftCrypt profiles.
• Amino acid code integration: Optionally incorporates amino acid code information to refine DTW-based alignments.
• Input format support: Processes chemical shift inputs formatted in NMR Exchange Format (NEF) and NMR-STAR.
• Sensitivity to disordered regions: Captures biophysical signals useful for analysis of intrinsically disordered protein regions.

Scientific Applications:

• Protein dynamics analysis: Interprets residue-level dynamics and conformational states from NMR chemical shift datasets.
• Biophysical comparison and alignment: Enables pairwise comparison of proteins based on biophysical profiles rather than sequence similarity alone.
• Analysis of intrinsically disordered regions: Facilitates characterization and comparison of disordered protein regions that are challenging for structure-based methods.

Methodology:

Processes NEF and NMR-STAR chemical shift inputs with a neural network auto-encoder to generate per-residue ShiftCrypt values, and applies Dynamic Time Warping (DTW), optionally with amino acid code information, for pairwise biophysical alignment.