Geometric Classification of TCM Formulas via Lo-Shu QLTE Scores from Real Crystallographic PDB Structures: Yin-Yang Partition and ChEMBL Validation

Key Points

• This research aims to classify traditional Chinese medicine (TCM) formulas using geometric scores derived from protein structures.
• Applied Lo-Shu Local Transition Energy (Q_LTE) scores computed from 41 crystallographic protein structures.
• Classified 8 TCM formulas according to 6 therapeutic mechanisms based on Q_LTE values.
• Conducted validation against 20 ChEMBL datasets to assess formula accuracy.
• Identified significant geometric partitioning: Huoxue targets in the Yin zone (Q_LTE<5.5) and HPA/immune/metabolic targets in the Yang zone (Q_LTE>5.5), with p=1.62e-5, Cohen d=-5.1.
• Significant discrimination for Huoxue formulas, Xuefu Zhuyu Tang (p=0.026, d=-1.44) and Buyang Huanwu Tang (p=0.018, d=-1.55).
• Achieved independent validation with Spearman rho=0.456 (p=0.044) against ChEMBL binding datasets.