ERα Virtual Screening of Ayurvedic Phytochemicals

# Virtual Screening of Ayurvedic Phytochemicals against Estrogen Receptor Alpha (ERα) ## Target **Estrogen Receptor Alpha (ERα)** — a nuclear receptor and transcription factor (UniProt: P03372) implicated in breast cancer, osteoporosis, and cardiovascular disease. The ligand-binding domain (LBD) is a validated drug target for selective estrogen receptor modulators (SERMs) and phytoestrogens. ## Workflow 1. **Receptor Acquisition** — Query RCSB PDB for the highest-resolution wild-type human ERα LBD crystal structure 2. **Receptor Preparation** — Remove waters/ligands, add hydrogens, assign charges → PDBQT 3. **Compound Library** — Assemble ~500 phytochemicals from IMPPAT 2.0, PubChem, and ChEMBL (Ayurvedic medicinal plants) 4. **Ligand Preparation** — 3D conformer generation (ETKDGv3 + MMFF94) → PDBQT via Meeko 5. **Pocket Detection** — Co-crystallized ligand centroid + ProDy validation → grid box 6. **Molecular Docking** — **GNINA** (CNN-scoring, primary) **+ AutoDock Vina** (comparison), exhaustiveness=8, 9 modes, 8 CPUs 7. **Docking Comparison** — Spearman rank correlation, score distributions, consensus hits (top-20 overlap) 8. **Interaction Diagrams** — 2D interaction maps for top 20 consensus compounds 9. **ADMET Profiling** — Lipinski Ro5, Veber, QED, PAINS, GI absorption, BBB permeation 10. **Report** — CSV exports, plots, and self-contained HTML report ## Data Sources | Source | Role | |--------|------| | RCSB PDB | ERα crystal structure | | IMPPAT 2.0 | Primary ph…

# ============================================================
# Cell 1a: Install System Libraries & Python Packages
# ============================================================
import subprocess, sys

print("=== Installing system libraries ===")
try:
    subprocess.run(["apt-get", "update", "-qq"],
                   stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, timeout=60)
    subprocess.run(["apt-get", "install", "-y", "-qq",
                    "libxrender1", "libxext6", "libx11-6"],
                   stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, timeout=60)
    print("  ✓ System libraries (libXrender, libXext, libX11)")
except Exception as e:
    print(f"  ⚠ System libs: {e}")

print("\n=== Installing Python packages ===")
for pkg in ["rdkit", "gemmi", "meeko", "vina", "biopandas",
…

=== Installing system libraries ===
  ✓ System libraries (libXrender, libXext, libX11)

=== Installing Python packages ===
  rdkit... ✓
  gemmi... ✓
  meeko... ✓
  vina... ✓
  biopandas... ✓
  prody... ✓
  openbabel-wheel... ✓
  pillow... ✓

✓ All Python packages installed.

# ============================================================
# Cell 1b: Download GNINA Binary (CPU, v1.1) & Verify
# ============================================================
import os, stat, subprocess, urllib.request

GNINA_BIN = "/usr/local/bin/gnina"
GNINA_URL = "https://github.com/gnina/gnina/releases/download/v1.1/gnina"

if not os.path.exists(GNINA_BIN):
    print(f"Downloading GNINA v1.1 (CPU, ~292 MB) ...")
    print(f"  URL: {GNINA_URL}")
    urllib.request.urlretrieve(GNINA_URL, GNINA_BIN)
    os.chmod(GNINA_BIN, os.stat(GNINA_BIN).st_mode | stat.S_IEXEC | stat.S_IXGRP | stat.S_IXOTH)
    size_mb = os.path.getsize(GNINA_BIN) / 1024 / 1024
    print(f"  ✓ Saved to {GNINA_BIN} ({size_mb:.0f} MB)")
else:
    size_mb = os.path.getsize(GNINA_BIN) / 1024 / 1024
    print(f"  ✓ GNINA already present ({size_mb:.0f} MB)")
…

Downloading GNINA v1.1 (CPU, ~292 MB) ...
  URL: https://github.com/gnina/gnina/releases/download/v1.1/gnina
  ✓ Saved to /usr/local/bin/gnina (292 MB)
  ✓ GNINA verified: Input:

# ============================================================
# Cell 1c: Import Libraries & Create Directory Structure
# ============================================================
import numpy as np
import pandas as pd
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import seaborn as sns

from rdkit import Chem, rdBase
from rdkit.Chem import (
    AllChem, Descriptors, Draw, QED, FilterCatalog,
    rdMolDescriptors, Lipinski, rdDistGeom, rdForceFieldHelpers
)
from rdkit.Chem import PandasTools
from rdkit.Chem.Draw import rdMolDraw2D

…

RDKit:         2026.03.4
Pandas:        2.2.3
NumPy:         2.1.2
ProDy:         2.6.1
Vina:          Python API ✓
GNINA:         /usr/local/bin/gnina ✓
Output dir:    /home/user/vs_results
Directories:   ['receptor', 'ligands_sdf', 'ligands_pdbqt', 'docking_vina', 'docking_gnina', 'interactions', 'plots', 'reports']

✓ All imports verified. Environment ready.

# ============================================================
# Cell 2: Download Highest-Quality Human ERα Structure from PDB
# ============================================================
import requests, json

print("=== Querying RCSB PDB for human ERα LBD structures ===\n")

# RCSB Search API v2 query
query = {
    "query": {
        "type": "group",
        "logical_operator": "and",
        "nodes": [
            {
                "type": "terminal",
                "service": "text",
                "parameters": {
                    "attribute": "rcsb_polymer_entity_container_identifiers.reference_sequence_identifiers.database_accession",
…

=== Querying RCSB PDB for human ERα LBD structures ===

Found 469 ERα X-ray structures (resolution ≤ 3.0 Å, ligand-bound)
Showing top 20 by resolution:

  7BAA | 1.10 Å | unknown | mut: WT
  8BZC | 1.10 Å | unknown | mut: WT
  8BZW | 1.10 Å | unknown | mut: WT
  8C04 | 1.10 Å | unknown | mut: WT
  7B9R | 1.15 Å | unknown | mut: WT
  7B9T | 1.15 Å | unknown | mut: WT
  6HMU | 1.20 Å | unknown | mut: C38N, S45C
  7BA8 | 1.20 Å | unknown | mut: WT
  8APS | 1.20 Å | unknown | mut: WT
  8BX3 | 1.20 Å…

# ============================================================
# Cell 3: Build Ayurvedic Phytochemical Dataset (~500 compounds)
# ============================================================
import requests, time, random
from rdkit import Chem
from rdkit.Chem import Descriptors

print("=== Building Ayurvedic Phytochemical Library ===\n")
print("Strategy: Query PubChem by compound name → guaranteed valid SMILES\n")

# ── Curated list of Ayurvedic phytochemical names ──────────
COMPOUND_NAMES = {
    "Curcuma longa": [
        "curcumin", "demethoxycurcumin", "bisdemethoxycurcumin",
        "ar-turmerone", "turmerone", "cyclocurcumin", "curcumenol",
        "curdione", "germacrone", "zederone",
    ],
    "Withania somnifera": [
…

=== Building Ayurvedic Phytochemical Library ===

Strategy: Query PubChem by compound name → guaranteed valid SMILES

Total compound names: 270 from 33 groups

Querying PubChem PUG REST...
  [50/270] found=50, failed=0
  [100/270] found=98, failed=2
  [150/270] found=146, failed=4
  [200/270] found=192, failed=8
  [250/270] found=242, failed=8
  [270/270] found=262, failed=8

  PubChem: 262 found, 8 not found

=== Deduplicating and filtering ===
  After dedup + MW filter: 250 unique compounds

=…

# ============================================================
# Cell 3b: Expand Library to ~500 Compounds (Supplementary Query)
# ============================================================
import requests, time, random
from rdkit import Chem
from rdkit.Chem import Descriptors

print("=== Expanding library with additional compounds ===\n")

# Additional compound names to reach ~500
ADDITIONAL_NAMES = {
    "Curcuma longa (extra)": [
        "bisacurone", "curcumol", "procurcumenol", "isoprocurcumenol",
        "curcumenone", "curcumanolide A", "curcumanolide B",
        "curcumanolide C", "curcumanolide D", "curcumanolide E",
    ],
    "Withania somnifera (extra)": [
        "withanolide F", "withanolide G", "withanolide H",
…