A Structure-Based Virtual Screening Protocol Utilizing PyPLIF HIPPOS and Vina for Targeting BACE-1

Alzheimer’s Disease (AD) is a leading cause of dementia, characterized by progressive cognitive decline driven in part by amyloid-β (Aβ) accumulation. The β-site amyloid precursor protein cleaving enzyme 1 (BACE-1) is responsible for initiating Aβ generation, making it a central therapeutic target. Yet, developing effective BACE-1 inhibitors has proven difficult due to structural complexity and pharmacological limitation. This study aimed to construct and validate a structure-based virtual screening (SBVS) workflow combining Vina and PyPLIF HIPPOS to facilitate the identification of promising BACE-1 ligands. The protocol was validated through 100 independent redocking experiments of the native ligand (PDB ID: 3L5F), all reproducing the crystallographic pose with RMSD < 2.0 Å. Large-scale screening of the DUDE dataset (283 active ligands; 18,100 decoys) generated interaction fingerprint, which were subsequently analyzed using Recursive Partitioning and Regression Trees (RPART) under varying prior probabilities. At the optimal prior ratio of 0.82:0.18, the model achieved an enrichment factor (EF) of 10.03, surpassing the DUDE benchmark (EF = 8.1). analysis consistently highlighted ionic interactions with Asp289 and hydrophobic contacts with Trp137 as key determinants of ligand activity. From 283 active ligands, 32 were classified as true positives, narrowing the pool of candidates and interpretable SBVS protocol and proposing a dual anchoring strategy involving Asp289 and Trp137 as a rational design principle for novel BACE-1 inhibitors in AD therapy.