MANORAA

MANORAA analyzes protein–ligand interactions and predicts the effects of spatial changes on binding affinity to guide structure-based ligand design using mapped structural data and machine learning.

Key Features:

• Structure-Based Analysis: Analyzes active-site pocket structures by examining pocket geometry, frequently occurring atoms, interatomic distances, and pocket boundaries.
• Grid-Based Mapping: Maps tens of thousands of amino acids onto a grid system to derive spatial features for model building.
• Predictive Algorithms: Generates model equations and uses predictive algorithms to estimate effects of contraction, expansion, and other spatial configuration changes on binding affinity.
• Machine Learning Techniques: Applies machine learning to structural data for model generation and prediction.
• Empirical Validation: Validates predictive capabilities with kinetic studies on dihydrofolate reductase (DHFR) and analyses of two DHFR-TS crystal structures.
• Extensive Structural Database: Interprets protein–ligand binding affinities through empirical analyses of 881 crystal structures involving 180 ligands.
• Integration with Biological Databases: Integrates structural analyses with major biological databases to contextualize results within existing drug-design resources.
• Application to Viral Proteins: Analyzes atom-frequency patterns within main protease structures, including SARS-CoV-2, to support ligand and probe design for viral targets.

Scientific Applications:

• Protein–Ligand Design: Guides optimization of ligand structures to improve binding affinity using structural and predictive analyses.
• Affinity Prediction and Optimization: Predicts how spatial modifications such as contraction or expansion affect binding affinity to inform rational ligand modification.
• Enzyme Modulation Studies: Supports interpretation of kinetic and structural determinants for enzymes exemplified by DHFR and DHFR-TS.
• Antiviral Probe Design: Informs design of ligands and probes targeting viral proteins, including the SARS-CoV-2 main protease.

Methodology:

Maps amino acids onto a grid, analyzes pocket structures (atoms, distances, boundaries), derives model equations using predictive algorithms and machine learning from empirical analyses of 881 crystal structures (180 ligands), and validates predictions against kinetic studies on DHFR and DHFR-TS.