Cordax

Cordax maps amyloid-forming sequence space using machine learning and high-resolution structural data from the Protein Data Bank (PDB) to identify sequence determinants of amyloid propensity.

Key Features:

• PDB-derived structural data: Leverages high-resolution amyloid core structures from the Protein Data Bank (PDB) for sequence analysis.
• Machine learning: Applies machine learning approaches to explore amyloid-forming capabilities beyond established sequence biases.
• t-SNE clustering: Clusters sequences using t-Distributed Stochastic Neighbour Embedding (t-SNE) for dimensionality reduction and visualization.
• Sequence-space expansion: Identifies amyloid-capable sequences beyond traditional hydrophobic, beta-sheet-prone motifs and Q/N/Y-rich yeast prions.
• Aliphatic and charge profiling: Detects clusters characterized by lower aliphatic content and higher net charge.
• Secondary-structure tendencies: Identifies clusters with tendencies toward helical structures and disordered conformations.
• Amyloid–solubility uncoupling: Reveals sequence contexts where amyloid propensity is uncoupled from solubility.
• Surface-exposed compatibility: Highlights sequence motifs compatible with surface-exposed patches in globular proteins.
• Functional and phase behavior links: Identifies motifs relevant to functional amyloids and sequences involved in liquid-liquid phase transitions.

Scientific Applications:

• Structural biology of amyloids: Provides perspectives on sequence determinants and structural features of amyloid cores.
• Protein misfolding diseases: Informs investigation of sequence variants and motifs relevant to amyloid-associated pathologies.
• Functional amyloids: Aids identification of sequences that form biologically functional amyloids.
• Liquid-liquid phase separation: Supports study of sequences involved in liquid-liquid phase transitions and related biophysical phenomena.
• Sequence-space discovery: Expands the catalog of amyloid-compatible sequences for downstream experimental or computational study.

Methodology:

Uses high-resolution amyloid core structures from the Protein Data Bank (PDB) and applies machine learning, including clustering with t-Distributed Stochastic Neighbour Embedding (t-SNE) for dimensionality reduction and visualization.