python3 << 'EOF' import pdfplumber import os files = { "qsar": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/qsar.pdf", "prodrug": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/pro drug.pdf", "cadd": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/cadd.pdf", "combchem": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/combination chem.pdf" } os.makedirs("/tmp/workspace/notes_project", exist_ok=True) for name, path in files.items(): try: with pdfplumber.open(path) as pdf: text = "" for i, page in enumerate(pdf.pages): t = page.extract_text() if t: text += f"\n--- PAGE {i+1} ---\n" + t out_path = f"/tmp/workspace/notes_project/{name}_raw.txt" with open(out_path, "w", encoding="utf-8") as f: f.write(text) print(f"{name}: {len(pdf.pages)} pages, {len(text)} chars -> {out_path}") except Exception as e: print(f"{name}: ERROR - {e}") EOF

python3 << 'EOF' from pypdf import PdfReader files = { "qsar": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/qsar.pdf", "prodrug": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/pro drug.pdf", "cadd": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/cadd.pdf", "combchem": "/tmp/workspace/attachments/3f0e76c5-2364-44f1-912f-06a1125a6b93/combination chem.pdf" } for name, path in files.items(): reader = PdfReader(path) text = "" for page in reader.pages: t = page.extract_text() if t: text += t + "\n" print(f"{name}: {len(text)} chars from pypdf") if text.strip(): with open(f"/tmp/workspace/notes_project/{name}_pypdf.txt","w") as f: f.write(text) EOF

python3 << 'PYEOF' from docx import Document from docx.shared import Pt, Inches, RGBColor from docx.enum.text import WD_ALIGN_PARAGRAPH import os os.makedirs("/tmp/workspace/notes_project", exist_ok=True) # ───────────────────────────────────────────────────────────────── # HELPER FUNCTIONS # ───────────────────────────────────────────────────────────────── def new_doc(title): doc = Document() # Normal style ns = doc.styles["Normal"] ns.font.name = "Arial" ns.font.size = Pt(11) # Title h = doc.add_heading(title, level=0) h.alignment = WD_ALIGN_PARAGRAPH.CENTER for run in h.runs: run.font.color.rgb = RGBColor(0x1F, 0x49, 0x7D) doc.add_paragraph() return doc def h1(doc, text): p = doc.add_heading(text, level=1) for run in p.runs: run.font.color.rgb = RGBColor(0x1F, 0x49, 0x7D) return p def h2(doc, text): p = doc.add_heading(text, level=2) for run in p.runs: run.font.color.rgb = RGBColor(0xC5, 0x50, 0x00) return p def h3(doc, text): return doc.add_heading(text, level=3) def body(doc, text): return doc.add_paragraph(text) def bullet(doc, text, level=0): p = doc.add_paragraph(text, style="List Bullet") p.paragraph_format.left_indent = Inches(0.25 * (level + 1)) return p def code_block(doc, text): """Monospace block for chemical structures/formulas""" p = doc.add_paragraph() run = p.add_run(text) run.font.name = "Courier New" run.font.size = Pt(9) p.paragraph_format.left_indent = Inches(0.4) # light grey shading via XML from docx.oxml.ns import qn from docx.oxml import OxmlElement pPr = p._p.get_or_add_pPr() shd = OxmlElement("w:shd") shd.set(qn("w:val"), "clear") shd.set(qn("w:color"), "auto") shd.set(qn("w:fill"), "F2F2F2") pPr.append(shd) return p def divider(doc): p = doc.add_paragraph("─" * 80) p.runs[0].font.color.rgb = RGBColor(0xCC, 0xCC, 0xCC) def add_table(doc, headers, rows): table = doc.add_table(rows=1+len(rows), cols=len(headers)) table.style = "Light Shading Accent 1" hdr = table.rows[0].cells for i, h in enumerate(headers): hdr[i].text = h for run in hdr[i].paragraphs[0].runs: run.font.bold = True for ri, row in enumerate(rows): cells = table.rows[ri+1].cells for ci, val in enumerate(row): cells[ci].text = val doc.add_paragraph() # ================================================================= # DOCUMENT 1 – QSAR # ================================================================= doc = new_doc("QSAR – Quantitative Structure Activity Relationship\nDeep-Dive Study Notes") h1(doc, "1. INTRODUCTION TO QSAR") body(doc, "QSAR is a mathematical/statistical approach that correlates the physicochemical, electronic, and structural properties of a molecule (descriptors) with its biological activity. The fundamental assumption is: similar structures have similar biological activities.") bullet(doc, "Developed by Corwin Hansch and Toshio Fujita (1964) – Hansch Analysis") bullet(doc, "Based on the Free-Wilson model (1964) – additive model") bullet(doc, "Goal: Predict biological activity of new/untested compounds without synthesis") bullet(doc, "Applications: Drug design, toxicity prediction, environmental risk assessment") h1(doc, "2. PHYSICOCHEMICAL PARAMETERS (DESCRIPTORS)") h2(doc, "2.1 Lipophilicity – Partition Coefficient (log P)") body(doc, "log P = log [Concentration in octanol] / [Concentration in water]") code_block(doc, "log P = log (C_octanol / C_water)\n\nHansch Equation (parabolic):\nlog (1/C) = a(log P) – b(log P)² + c\n\nwhere:\n C = molar concentration for standard biological response\n a,b,c = regression coefficients\n Optimal log P (π₀) = a / 2b") bullet(doc, "log P > 0 → lipophilic; log P < 0 → hydrophilic") bullet(doc, "Optimal log P for CNS drugs ≈ 2") bullet(doc, "Optimal log P for oral drugs ≈ 1–3 (Lipinski: ≤ 5)") bullet(doc, "Hydrophobic substituent constant π (pi) = log P_X – log P_H (Hansch)") body(doc, "Substituent π values:") code_block(doc, "Substituent π value\n─────────────────────────\n–H 0.00\n–CH₃ +0.56\n–C₂H₅ +1.02\n–Cl +0.71\n–F +0.14\n–OH –0.67\n–NH₂ –1.23\n–NO₂ –0.28\n–COOH –0.32") h2(doc, "2.2 Electronic Parameters") body(doc, "Hammett Sigma (σ) – measures electron withdrawing/donating effect of substituent") code_block(doc, "Hammett Equation:\nlog (K/K₀) = ρ · σ\n\nwhere:\n K = ionisation constant of substituted benzoic acid\n K₀ = ionisation constant of benzoic acid\n ρ = reaction constant (sensitivity of reaction to electronic effects)\n σ = substituent constant\n\nSigma values:\nSubstituent σ_para σ_meta\n──────────────────────────────────\n–H 0.00 0.00\n–NO₂ +0.78 +0.71\n–CN +0.66 +0.56\n–Cl +0.23 +0.37\n–F +0.06 +0.34\n–CH₃ –0.17 –0.07\n–OH –0.37 +0.12\n–NH₂ –0.66 –0.16\n–OCH₃ –0.27 +0.12") bullet(doc, "σ > 0 → electron withdrawing (EWG)") bullet(doc, "σ < 0 → electron donating (EDG)") bullet(doc, "σ_para vs σ_meta: para includes resonance + inductive; meta = inductive only") h2(doc, "2.3 Steric Parameters") body(doc, "Taft Steric Parameter (Es): measures steric effect of substituents") code_block(doc, "Es = log (k_s / k_CH₃)\n\nwhere k_s = rate constant for substituted ester hydrolysis\n k_CH₃ = rate constant for methyl ester\n\nTaft Es values:\nSubstituent Es\n──────────────────\n–H +1.24\n–CH₃ 0.00\n–C₂H₅ –0.07\n–i-Pr –0.47\n–t-Bu –1.54\n–Ph –2.55") body(doc, "Molar Refractivity (MR): combines polarisability and volume") code_block(doc, "MR = (n² – 1)/(n² + 2) × MW/d\n\nwhere n = refractive index, MW = molecular weight, d = density\nMR relates to London dispersion forces (van der Waals)") body(doc, "Verloop STERIMOL parameters: L, B1, B2, B3, B4, B5 – capture shape/directionality of substituents") h1(doc, "3. HANSCH ANALYSIS (Linear Free Energy Relationship)") body(doc, "General Hansch equation:") code_block(doc, "log (1/C) = k₁ · π – k₂ · π² + k₃ · σ + k₄ · MR + k₅\n\nOr simplified:\nlog (1/C) = a · log P + b · σ + c · Es + d\n\nParabolic form used when activity peaks at optimal hydrophobicity:\nlog (1/C) = –k(log P – log P₀)² + k₃σ + k₄") bullet(doc, "Requires minimum 5–6 compounds per variable in training set") bullet(doc, "Uses multiple linear regression (MLR)") bullet(doc, "Validates with r² > 0.9, cross-validation (q²)") bullet(doc, "Predictive power assessed by test set or leave-one-out (LOO)") h2(doc, "3.1 Craig Plot") body(doc, "2D plot of σ (x-axis) vs π (y-axis) dividing into 4 quadrants:") code_block(doc, " π (+)\n │\n II │ I\n EDG + │ EWG +\n lipophil│ lipophil\n─────────┼─────────→ σ\n III │ IV\n EDG – │ EWG –\n hydrophil│ hydrophil\n │\n\nQuadrant I: EWG + lipophilic (e.g. –Cl, –Br, –CF₃)\nQuadrant II: EDG + lipophilic (e.g. –CH₃, –C₂H₅)\nQuadrant III: EDG + hydrophilic (e.g. –OH, –NH₂)\nQuadrant IV: EWG + hydrophilic (e.g. –NO₂, –COOH)") bullet(doc, "Craig plot helps in bioisosteric replacement and lead optimisation") h2(doc, "3.2 Topliss Scheme (Operational Scheme)") body(doc, "Manual QSAR method for rapid analog synthesis and testing without regression analysis") code_block(doc, "Topliss Tree for Aromatic Substitution:\n\n Start: unsubstituted compound\n ↓\n Test 4-Cl derivative\n / \\\n More active Less active\n ↓ ↓\n Test 4-NO₂ Test 4-OMe\n Test 3,4-Cl₂ (if less → EDG unfav)\n\nConclusion from activity trend → decide next substituent\nMM = Most active, L = Less active, E = Equal") h1(doc, "4. FREE–WILSON ANALYSIS") body(doc, "Additive model: each structural fragment contributes independently to activity") code_block(doc, "log BA = Σ aᵢ · Xᵢ + μ\n\nwhere:\n BA = biological activity\n aᵢ = contribution of substituent i\n Xᵢ = indicator variable (1 if present, 0 if absent)\n μ = overall average activity (parent compound)") bullet(doc, "Pure structure-based; no physicochemical data needed") bullet(doc, "Limitation: cannot extrapolate beyond substituents tested") bullet(doc, "Combined Hansch-Free Wilson: mixed model for better predictive power") h1(doc, "5. 3D-QSAR METHODS") h2(doc, "5.1 CoMFA – Comparative Molecular Field Analysis") body(doc, "Aligns molecules in 3D space, calculates steric (Lennard-Jones) and electrostatic (Coulombic) fields at grid points") code_block(doc, "Steric field: E_steric = 4ε[(σ/r)¹² – (σ/r)⁶] (Lennard-Jones 6-12)\nElectrostatic: E_elec = q₁q₂ / ε·r (Coulomb's law)\n\nPLS (Partial Least Squares) regression:\nbio activity = f(steric field values, electrostatic field values)") bullet(doc, "Output: 3D contour maps showing regions where bulky/electron-rich groups increase or decrease activity") bullet(doc, "Green contours: bulk tolerance (add steric bulk here)") bullet(doc, "Yellow contours: bulk disfavoured (avoid steric bulk)") bullet(doc, "Blue contours: positive charge favoured") bullet(doc, "Red contours: negative charge favoured") h2(doc, "5.2 CoMSIA – Comparative Molecular Similarity Indices Analysis") body(doc, "Extension of CoMFA; uses Gaussian functions; includes steric, electrostatic, hydrophobic, H-bond donor, H-bond acceptor fields") h2(doc, "5.3 Pharmacophore Modeling") bullet(doc, "Identifies minimum essential features (H-bond donor/acceptor, hydrophobic centre, +/– charge, aromatic ring)") bullet(doc, "Tools: CATALYST, Phase, LigandScout") h1(doc, "6. MOLECULAR DESCRIPTORS OVERVIEW") body(doc, "Types of descriptors:") add_table(doc, ["Type", "Examples", "Software"], [ ["Constitutional", "MW, atom count, ring count", "Dragon"], ["Topological", "Wiener index, Zagreb index, Randic index", "Dragon, Padel"], ["Geometrical (3D)", "Shadow indices, WHIM, GETAWAY", "Dragon"], ["Electronic", "Partial charges, dipole moment, HOMO/LUMO", "Gaussian, MOE"], ["Pharmacophoric", "H-bond, charge centres", "Catalyst, Phase"], ] ) h1(doc, "7. LIPINSKI'S RULE OF FIVE (Ro5)") body(doc, "Predicts oral bioavailability (drug-likeness):") code_block(doc, "1. MW ≤ 500 Da\n2. log P ≤ 5 (lipophilicity)\n3. H-bond donors ≤ 5 (–NH, –OH)\n4. H-bond acceptors ≤ 10 (N, O atoms)\n5. Rotatable bonds ≤ 10 (Veber rule)\n6. PSA (Polar Surface Area) ≤ 140 Ų (Veber)\n\nViolations > 1 → poor oral bioavailability\nExceptions: substrates of active transporters (e.g., antibiotics)") h1(doc, "8. QSAR VALIDATION PARAMETERS") code_block(doc, "r² = coefficient of determination (fitting quality, should be > 0.9)\nq² = cross-validated r² (predictive power, should be > 0.5)\nF = Fisher's test (statistical significance)\ns = standard error of estimate\nrms = root mean square error\nLOO = Leave-One-Out cross validation\nExternal test set validation: r²_pred > 0.6") h1(doc, "9. APPLICATIONS AND LIMITATIONS") h2(doc, "Applications") bullet(doc, "Lead optimisation and analog design") bullet(doc, "Prediction of ADMET properties") bullet(doc, "Virtual screening of compound libraries") bullet(doc, "Metabolic stability prediction") bullet(doc, "Toxicity prediction (QSTR)") h2(doc, "Limitations") bullet(doc, "Dependent on quality and size of training set") bullet(doc, "Cannot account for receptor flexibility") bullet(doc, "Assumes linearity of contributions") bullet(doc, "3D methods require correct molecular alignment") bullet(doc, "Overfitting risk with too many descriptors") doc.save("/tmp/workspace/notes_project/1_QSAR_Notes.docx") print("QSAR done") # ================================================================= # DOCUMENT 2 – PRO DRUG # ================================================================= doc = new_doc("PRO DRUGS – Concept, Design & Classification\nDeep-Dive Study Notes") h1(doc, "1. DEFINITION AND CONCEPT") body(doc, "A prodrug is a pharmacologically inactive (or less active) compound that is converted (bioactivated) to the active drug (parent drug) inside the body by enzymatic or chemical means.") bullet(doc, "Coined by: Adrian Albert (1958)") bullet(doc, "Also called: Pro-agent, Bioreversible derivative, Latentiated drug") bullet(doc, "Active drug released after metabolic transformation") bullet(doc, "Opposite of a prodrug = soft drug (active compound metabolised to inactive metabolite)") code_block(doc, "PRODRUG ──[Bioactivation]──→ ACTIVE DRUG ──[Action]──→ Pharmacological Effect\n (inactive) (enzymes/pH) (parent drug)") h1(doc, "2. RATIONALE / NEED FOR PRODRUGS") body(doc, "Problems with parent drug that necessitate prodrug design:") bullet(doc, "Poor aqueous solubility (formulation problems)") bullet(doc, "Poor oral bioavailability (first-pass metabolism, P-gp efflux)") bullet(doc, "Unpleasant taste/odour (patient compliance)") bullet(doc, "Local irritation/toxicity at site of administration") bullet(doc, "Short duration of action (rapid metabolism/excretion)") bullet(doc, "Poor permeability across biological membranes (BBB, GI tract)") bullet(doc, "Lack of site-specificity (systemic side effects)") bullet(doc, "Chemical instability of parent drug") h1(doc, "3. CLASSIFICATION OF PRODRUGS") h2(doc, "Type I Prodrugs – Intracellular Bioactivation") body(doc, "Activated inside cells by intracellular enzymes") bullet(doc, "Type Ia: Bioactivated in metabolically active cells (therapeutic target cells)") bullet(doc, "Example: Antivirals – AZT (Zidovudine), Acyclovir") code_block(doc, "Acyclovir (prodrug) ──[Viral thymidine kinase]──→ Acyclovir monophosphate\n ──[Cellular kinases]──────→ Acyclovir triphosphate (ACTIVE)\n\nAcyclovir structure:\n O\n ‖\n HN─C─NH\n / \\\n N N\n ‖ │\n C CH₂─O─CH₂─CH₂─OH\n \\ /\n C═════C\n │\n (guanine base)") bullet(doc, "Type Ib: Bioactivated in cells not at therapeutic target (remote cells)") bullet(doc, "Example: L-DOPA (Levodopa) → Dopamine in brain") code_block(doc, "Levodopa (L-DOPA): Dopamine (Active):\n HOOC─CH─CH₂ HO─ ─CH₂─CH₂─NH₂\n │ \\_/\n NH₂ (3,4-dihydroxyphenylethylamine)\n HO─ ─OH\n \\─/\n (catechol ring)\n\nConversion: DOPA decarboxylase\nL-DOPA crosses BBB; dopamine cannot (too polar)") h2(doc, "Type II Prodrugs – Extracellular Bioactivation") body(doc, "Activated in GI fluids, blood, lymph, or interstitial fluid") bullet(doc, "Type IIa: Activated in GI tract (gut lumen, gut wall)") bullet(doc, "Example: Aspirin (acetylsalicylic acid) – hydrolysed to salicylic acid") bullet(doc, "Example: Bacampicillin → Ampicillin (esterase in gut wall)") code_block(doc, "Bacampicillin: Ampicillin (Active):\n Ampicillin─C(=O)─O─CH(CH₃)─O─C(=O)─OEt\n ↓ [esterase]\n Ampicillin + acetaldehyde + CO₂ + EtOH\n\nAdvantage: Better oral bioavailability (98% vs 40% for ampicillin)") bullet(doc, "Type IIb: Activated systemically (plasma, blood enzymes)") bullet(doc, "Example: Enalapril → Enalaprilat (ACE inhibitor)") code_block(doc, "Enalapril (Prodrug): Enalaprilat (Active):\n –COOC₂H₅ (ethyl ester) –COOH (free acid)\n ↓ [plasma esterase] ↑\n Oral bioavailability Active ACE inhibitor\n much better than enalaprilat (too polar to absorb orally)") h1(doc, "4. CHEMICAL APPROACHES TO PRODRUG DESIGN") h2(doc, "4.1 Ester Prodrugs") body(doc, "Most common approach; ester hydrolysed by esterases to release free acid or alcohol") code_block(doc, "R─COOH + HO─R' → R─COO─R' + H₂O\n(active acid) (ester prodrug)\n\nExamples:\n• Aspirin: –OH esterified to –OOCCH₃ (released as salicylate)\n• Enalapril: –COOH ethyl ester (released as enalaprilat)\n• Chloramphenicol palmitate: –OH esterified (tasteless prodrug)\n• Clindamycin phosphate: phosphate ester (better solubility)\n• Pivampicillin: pivaloyloxymethyl ester of ampicillin") h2(doc, "4.2 Amide and Carbamate Prodrugs") code_block(doc, "R─NH₂ + R'─COOH → R─NH─CO─R' (amide prodrug)\n\nExample: Pivmecillinam (amide prodrug of mecillinam)\n\nCarbamate: R─OH → R─O─CO─NHR'\n Hydrolysed by carbamylases/esterases") h2(doc, "4.3 Phosphate Prodrugs") code_block(doc, "R─OH → R─O─PO₃H₂\n\nExamples:\n• Prednisolone phosphate (water soluble → prednisolone for IV)\n• Fosphenytoin → Phenytoin (water soluble phosphate ester)\n• Fosamprenavir → Amprenavir\n• Tenofovir disoproxil fumarate → Tenofovir\n\nAdvantage: Dramatically increases aqueous solubility\nCleavage: by alkaline phosphatase in intestine/blood") h2(doc, "4.4 Double Prodrugs (Pro-Prodrugs)") code_block(doc, "Pro-Prodrug ──[step 1]──→ Prodrug ──[step 2]──→ Active Drug\n\nExample: Pivampicillin:\n Ampicillin─OOCCH₂OC(=O)CMe₃\n ↓ [esterase – step 1]\n Ampicillin─OOCCH₂OH (unstable hemiacetal)\n ↓ [spontaneous – step 2]\n Ampicillin─COOH + CH₂O\n (Active ampicillin)") h2(doc, "4.5 Mutual Prodrugs (Codrugs)") body(doc, "Two pharmacologically active compounds linked together; one acts as prodrug carrier for the other") code_block(doc, "Drug A ─ Linker ─ Drug B\n\nExample: Sultamicillin = Ampicillin─O─C(=O)─O─Sulbactam\n (linked through carbonate ester)\n Both released simultaneously on hydrolysis\n Synergistic antibacterial effect") h2(doc, "4.6 Site-Specific / Targeted Prodrugs") code_block(doc, "ADEPT – Antibody-Directed Enzyme Prodrug Therapy:\n Antibody─Enzyme conjugate targets tumour\n Prodrug administered → Enzyme at tumour activates prodrug → Active drug released locally\n\nGDEPT – Gene-Directed Enzyme Prodrug Therapy:\n Gene for activating enzyme inserted into tumour cells\n Prodrug administered → activated only in tumour\n\nExample: 5-Fluorocytosine → 5-Fluorouracil\n (via cytosine deaminase expressed in tumour)") h1(doc, "5. BIOREVERSIBLE DERIVATIVES (Carrier-linked Prodrugs)") h2(doc, "5.1 Bipartite Prodrugs") body(doc, "Drug + carrier (promoiety) linked by a single bond that is cleaved in vivo") code_block(doc, "Structure: Drug─[bond]─Carrier\nCarrier = promoiety (non-toxic, rapidly excreted)\nBond type: ester, amide, carbamate, carbonate, disulphide") h2(doc, "5.2 Tripartite Prodrugs") body(doc, "Drug ─ linker ─ carrier; linker provides controlled release") code_block(doc, "Structure: Drug─[linker]─Carrier\n\nExample: Drug─O─CH₂─O─Carrier\n(Acyloxymethyl type linker – undergoes spontaneous cyclisation)") h1(doc, "6. IMPORTANT PRODRUG EXAMPLES (Exam-Ready)") add_table(doc, ["Prodrug", "Active Drug", "Mechanism", "Advantage"], [ ["Levodopa (L-DOPA)", "Dopamine", "DOPA decarboxylase in brain", "BBB penetration"], ["Enalapril", "Enalaprilat", "Plasma esterase", "Oral bioavailability"], ["Bacampicillin", "Ampicillin", "Gut wall esterase", "Better oral absorption"], ["Pivampicillin", "Ampicillin", "Double ester hydrolysis", "Enhanced absorption"], ["Acyclovir", "Acyclovir-TP", "Viral + cellular kinases", "Selective antiviral"], ["Valacyclovir", "Acyclovir", "Intestinal esterase", "Oral bioavailability 54% vs 20%"], ["Fosphenytoin", "Phenytoin", "Phosphatase", "Water soluble IV form"], ["Codeine", "Morphine", "CYP2D6 O-demethylation", "Oral analgesic"], ["Sulfasalazine", "5-Aminosalicylate", "Colonic bacteria azoreductase", "Colon targeting"], ["Chloramphenicol palmitate", "Chloramphenicol", "GI esterase", "Tasteless formulation"], ["Capecitabine", "5-Fluorouracil", "Thymidine phosphorylase (tumour)", "Tumour targeting"], ["Oseltamivir (Tamiflu)", "Oseltamivir carboxylate", "Hepatic esterase", "Oral bioavailability"], ] ) h1(doc, "7. EVALUATION OF PRODRUGS") bullet(doc, "In vitro hydrolysis studies (plasma, tissue homogenate)") bullet(doc, "Bioavailability comparison (prodrug vs parent drug, AUC, Cmax)") bullet(doc, "Metabolite identification (LC-MS/MS)") bullet(doc, "Toxicity of carrier/promoiety – must be non-toxic") bullet(doc, "Stability studies – must be stable during storage, absorb intact") bullet(doc, "Enzyme specificity studies – identify bioactivating enzyme") h1(doc, "8. ADVANTAGES AND DISADVANTAGES") h2(doc, "Advantages") bullet(doc, "Improved oral bioavailability and absorption") bullet(doc, "Increased duration of action (depot effects)") bullet(doc, "Reduced toxicity and side effects") bullet(doc, "Improved patient compliance (taste, odour, pain)") bullet(doc, "Targeted drug delivery (organ/tissue specificity)") bullet(doc, "Overcome first-pass metabolism") bullet(doc, "Improved water solubility for parenteral use") h2(doc, "Disadvantages") bullet(doc, "Unpredictable/variable bioactivation in patients") bullet(doc, "Possible toxic metabolites from carrier") bullet(doc, "Enzyme deficiency in some patients (e.g., CYP2D6 poor metabolisers for codeine)") bullet(doc, "Complex synthesis and higher cost") bullet(doc, "Regulatory complexity – two NCEs to characterise") doc.save("/tmp/workspace/notes_project/2_ProDrug_Notes.docx") print("ProDrug done") # ================================================================= # DOCUMENT 3 – CADD (Computer Aided Drug Design) # ================================================================= doc = new_doc("CADD – Computer Aided Drug Design\nDeep-Dive Study Notes") h1(doc, "1. INTRODUCTION TO CADD") body(doc, "CADD refers to the use of computational methods and tools to facilitate the discovery, design, and optimisation of drug molecules. It integrates molecular biology, chemistry, bioinformatics, and computational physics.") bullet(doc, "Reduces cost and time of drug discovery (traditional: 10–15 years, $1–2 billion)") bullet(doc, "Filters large virtual libraries before synthesis") bullet(doc, "Helps identify binding mode and key interactions") bullet(doc, "Two major categories: Structure-Based Drug Design (SBDD) and Ligand-Based Drug Design (LBDD)") code_block(doc, "Drug Discovery Pipeline with CADD:\n\n Target ID → Target Validation → Hit ID → Lead Optimisation → Pre-clinical\n ↓ ↓ ↓ ↓\n Bioinformatics Homology Virtual ADMET/Docking\n (protein Modelling Screening Optimisation\n structure)") h1(doc, "2. STRUCTURE-BASED DRUG DESIGN (SBDD)") body(doc, "Uses 3D structure of biological target (protein/enzyme/receptor) to design complementary ligands. Target structure obtained from X-ray crystallography, NMR, or Cryo-EM.") h2(doc, "2.1 Molecular Docking") body(doc, "Predicts the preferred binding orientation and conformation of a small molecule (ligand) within the binding site of a macromolecular target (receptor)") code_block(doc, "Steps in Molecular Docking:\n\n1. Prepare receptor: Remove water, add H atoms, assign charges (Gasteiger/AMBER)\n2. Define binding site: Grid box around active site\n3. Prepare ligand: Generate 3D structure, torsion angles, charges\n4. Docking algorithm: Search conformational space\n5. Scoring: Rank poses by binding affinity\n6. Post-processing: Cluster, visualise, validate\n\nKey Scoring Functions:\n• Force-field based: E = E_vdW + E_elec + E_H-bond\n• Empirical: ΔG_bind = ΔG_vdW + ΔG_elec + ΔG_HB + ΔG_desolv + ΔG_rot\n• Knowledge-based: Statistical potentials from PDB structures") body(doc, "Common Docking Software:") add_table(doc, ["Software", "Algorithm", "Scoring"], [ ["AutoDock 4", "Lamarckian Genetic Algorithm", "Semi-empirical free energy"], ["AutoDock Vina", "Iterated local search", "Hybrid empirical"], ["Glide (Schrödinger)", "Systematic/MCMM", "GlideScore (SP, XP)"], ["GOLD", "Genetic Algorithm", "ChemScore, GoldScore"], ["FlexX", "Incremental construction", "Böhm scoring"], ["DOCK", "Shape matching", "Force field + GB/SA"], ] ) h2(doc, "2.2 Binding Free Energy Calculation") code_block(doc, "ΔG_binding = G_complex – G_receptor – G_ligand\n\nMM-GBSA / MM-PBSA approach:\nΔG_bind = ΔH_MM + ΔG_solvation – TΔS\n\nFEP (Free Energy Perturbation):\nΔΔG = ΔG(A→B in complex) – ΔG(A→B in solution)\n\nThermodynamic Cycle:\n Ligand A (solution) ──ΔG_sol──→ Ligand B (solution)\n ↓ ↓\n ΔG_bind(A) ΔG_bind(B)\n ↓ ↓\n Complex A ──ΔG_complex→ Complex B") h2(doc, "2.3 De Novo Drug Design") body(doc, "Builds new molecules from scratch within the binding pocket") bullet(doc, "Fragment-based: assemble pharmacophoric fragments in active site") bullet(doc, "Tools: LUDI, GROW, LeapFrog, LigBuilder") bullet(doc, "Fragment Linking: link two fragments binding at adjacent sites") bullet(doc, "Fragment Growing: extend fragment at one site") code_block(doc, "FBDD (Fragment-Based Drug Discovery) workflow:\n\n Screen fragment library (MW ~150 Da, few pharmacophores)\n ↓ (SPR, NMR, X-ray)\n Identify weak-binding fragments (Kd ~mM)\n ↓ \n Fragment optimisation / linking\n ↓\n Lead compound (nM binders)\n\nExample: Vemurafenib (B-RAF inhibitor) – developed via FBDD") h1(doc, "3. LIGAND-BASED DRUG DESIGN (LBDD)") body(doc, "Used when 3D structure of target is NOT known. Uses known active ligands to extract structure-activity information.") h2(doc, "3.1 Pharmacophore Modelling") body(doc, "A pharmacophore is the ensemble of steric and electronic features necessary for optimal interaction with a specific biological target") code_block(doc, "Pharmacophoric features:\n• HBD – Hydrogen Bond Donor (e.g. –NH, –OH)\n• HBA – Hydrogen Bond Acceptor (e.g. C=O, –N:)\n• Hyd – Hydrophobic centre (lipophilic area)\n• Pos – Positive ionisable (–NH₃⁺, amidine)\n• Neg – Negative ionisable (–COO⁻, sulphonate)\n• Aro – Aromatic ring (π-π stacking)\n\nPharmacophore Development Steps:\n1. Collect set of active molecules (structurally diverse)\n2. Superimpose / align molecules\n3. Identify common 3D features with distances\n4. Build pharmacophore model\n5. Validate with known actives/inactives\n6. Virtual screening of databases (ZINC, ChEMBL)\n\nTools: CATALYST (Hypogen/Hiphop), Phase (Schrödinger),\n LigandScout, MOE Pharmacophore") h2(doc, "3.2 Similarity Searching") code_block(doc, "Tanimoto coefficient (Tc):\nTc = |A ∩ B| / |A ∪ B| = c / (a + b – c)\n\nwhere:\n a = bits set in molecule A\n b = bits set in molecule B\n c = bits set in both A and B\n\nTc = 1.0 → identical; Tc > 0.85 → very similar\n\nFingerprint types:\n• MACCS keys (166 structural keys)\n• ECFP (Extended Connectivity Fingerprints) / Morgan\n• Daylight fingerprints\n• FCFP (Feature-based)\n• RDKit fingerprints") h2(doc, "3.3 3D-QSAR (CoMFA, CoMSIA)") body(doc, "Already covered in QSAR notes – key CADD application") h1(doc, "4. VIRTUAL SCREENING (VS)") body(doc, "High-throughput computational screening of large compound databases to identify potential drug candidates") code_block(doc, "Virtual Screening Workflow:\n\n Large compound database (millions of compounds)\n e.g. ZINC, PubChem, ChEMBL, Enamine\n ↓\n Filter 1: Lipinski/drug-likeness filters\n ↓\n Filter 2: Pharmacophore screening\n ↓\n Filter 3: Docking (rigid receptor)\n ↓\n Filter 4: Rescoring / MM-GBSA\n ↓\n Hit list (100–1000 compounds)\n ↓\n Experimental validation (IC₅₀, Kd, cell assay)") h2(doc, "Types of Virtual Screening") bullet(doc, "SBVS (Structure-Based): molecular docking into target binding site") bullet(doc, "LBVS (Ligand-Based): pharmacophore, similarity, 3D-QSAR") bullet(doc, "Hybrid: combine both approaches") h1(doc, "5. HOMOLOGY MODELLING") body(doc, "Builds 3D structure of target protein from its amino acid sequence when experimental structure is unavailable, using a known structure of a homologous protein as template") code_block(doc, "Steps in Homology Modelling:\n\n1. Target sequence → BLAST search → Template selection\n2. Target-template alignment (ClustalW, MUSCLE)\n3. Backbone generation (using template Cα coordinates)\n4. Loop modelling (MODELLER, Rosetta)\n5. Side chain placement\n6. Energy minimisation (AMBER, CHARMM)\n7. Model validation:\n • Ramachandran plot (>90% residues in allowed regions)\n • PROCHECK, WHAT_CHECK, DOPE score\n • ProSA Z-score\n\nTools: MODELLER, SWISS-MODEL, I-TASSER, Phyre2\nTemplate identity: >30% → reliable model; <30% → use with caution") h1(doc, "6. MOLECULAR DYNAMICS (MD) SIMULATION") body(doc, "Simulates motion of atoms/molecules over time using Newton's equations of motion") code_block(doc, "Newton's equation: F = ma → d²r/dt² = F/m\n\nF = –∇V(r) (force = negative gradient of potential energy)\n\nPotential Energy Function (Force Field):\nV = Σ½k_b(r–r₀)² + Σ½k_θ(θ–θ₀)² + Σ½V_n[1+cos(nφ–γ)]\n bonds angles dihedrals\n + Σ[4ε((σ/r)¹²–(σ/r)⁶)] + Σ(q_iq_j/4πε₀r)\n van der Waals electrostatics\n\nForce Fields: AMBER, CHARMM, GROMOS, OPLS\nSoftware: GROMACS, AMBER, NAMD, CHARMM\n\nMD Applications:\n• Study protein flexibility / induced fit\n• Calculate binding free energies\n• Identify allosteric sites\n• Drug resistance mechanisms") h1(doc, "7. ADMET PREDICTION") body(doc, "In silico prediction of Absorption, Distribution, Metabolism, Excretion, Toxicity") add_table(doc, ["Property", "Key Parameters", "Tools"], [ ["Absorption", "Caco-2 permeability, P-gp substrate, Pgp inhibitor", "admetSAR, pkCSM"], ["Distribution", "log BB (brain/blood), PPB (plasma protein binding), Vd", "SwissADME"], ["Metabolism", "CYP450 substrate/inhibitor (1A2,2C9,2C19,2D6,3A4)", "StarDrop"], ["Excretion", "Renal clearance, half-life", "ADMET Predictor"], ["Toxicity", "hERG (cardiotox), Ames test (mutagenicity), LD50", "admetSAR, Derek"], ] ) code_block(doc, "Key ADMET descriptors:\n• PSA (Polar Surface Area) < 140 Ų → good oral absorption\n• PSA < 90 Ų → good BBB penetration\n• log P = 1–3 → optimal for oral drugs\n• HBD ≤ 5, HBA ≤ 10 (Lipinski Ro5)\n• MW < 500 Da (Lipinski)\n• pKa → ionisation state at physiological pH") h1(doc, "8. KEY DATABASES USED IN CADD") add_table(doc, ["Database", "Content", "URL"], [ ["PDB (Protein Data Bank)", "3D protein structures (X-ray, NMR, cryo-EM)", "rcsb.org"], ["ChEMBL", "Bioactivity data, drug targets", "ebi.ac.uk/chembl"], ["ZINC", "Commercially available compounds for VS", "zinc.docking.org"], ["DrugBank", "Drug-target interactions", "drugbank.ca"], ["PubChem", "Chemical compounds, bioassays", "pubchem.ncbi.nlm.nih.gov"], ["BindingDB", "Protein-ligand binding affinities", "bindingdb.org"], ["UniProt", "Protein sequences and function", "uniprot.org"], ] ) h1(doc, "9. IMPORTANT CADD SOFTWARE SUMMARY") add_table(doc, ["Task", "Software"], [ ["Molecular Docking", "AutoDock, AutoDock Vina, Glide, GOLD, FlexX"], ["MD Simulation", "GROMACS, AMBER, NAMD, CHARMM"], ["Homology Modelling", "MODELLER, SWISS-MODEL, I-TASSER"], ["Pharmacophore", "CATALYST, Phase, LigandScout, MOE"], ["QSAR/3D-QSAR", "SYBYL (CoMFA), Discovery Studio"], ["Visualisation", "PyMOL, Discovery Studio Visualizer, UCSF Chimera"], ["ADMET", "SwissADME, admetSAR, pkCSM, StarDrop"], ["Protein Prep", "Schrödinger Protein Prep Wizard"], ["Virtual Screening", "Vina, Glide, DOCK, Schrödinger VS"], ] ) h1(doc, "10. CADD IN DRUG DISCOVERY – NOTABLE EXAMPLES") bullet(doc, "HIV protease inhibitors (Saquinavir, Ritonavir): first rationally designed drugs using SBDD") bullet(doc, "Dorzolamide (carbonic anhydrase inhibitor): structure-based") bullet(doc, "Imatinib (Gleevec): designed using BCR-ABL crystal structure") bullet(doc, "Zanamivir (Relenza): structure-based neuraminidase inhibitor") bullet(doc, "Vemurafenib: fragment-based design targeting B-RAF V600E") bullet(doc, "Osimertinib: EGFR inhibitor – CADD guided resistance mutation targeting") doc.save("/tmp/workspace/notes_project/3_CADD_Notes.docx") print("CADD done") # ================================================================= # DOCUMENT 4 – COMBINATORIAL CHEMISTRY # ================================================================= doc = new_doc("COMBINATORIAL CHEMISTRY\nDeep-Dive Study Notes") h1(doc, "1. INTRODUCTION") body(doc, "Combinatorial chemistry is a technique by which a large number of structurally diverse small molecules (libraries) can be synthesised simultaneously using systematic and repetitive chemical synthesis. It accelerates drug discovery by rapidly generating large compound collections for biological screening.") bullet(doc, "Traditional synthesis: 1 compound at a time") bullet(doc, "Combinatorial synthesis: hundreds to millions of compounds simultaneously") bullet(doc, "Combined with High-Throughput Screening (HTS) → rapid hit identification") bullet(doc, "Emerged in 1990s; revolutionised pharmaceutical lead identification") bullet(doc, "Pioneers: R.B. Merrifield (solid-phase peptide synthesis, Nobel 1984), Geysen, Furka") h1(doc, "2. KEY CONCEPTS") h2(doc, "2.1 Library") body(doc, "A collection of compounds prepared by combinatorial methods, all sharing a common scaffold but varying substituents") code_block(doc, "Scaffold (Core) + Varying Building Blocks → Library\n\nExample: Benzodiazepine library\n\n R₁\n │\n N─C\n / \\\n C C═O\n │ │\n C NH\n \\ /\n C═C\n │\n R₂\n\nVary R₁ and R₂ → large benzodiazepine library") h2(doc, "2.2 Building Blocks (Monomers)") body(doc, "Reactant molecules that are combined in different combinations to generate library members. Must be commercially available, reactive, and structurally diverse.") h2(doc, "2.3 Scaffold / Core Structure") body(doc, "The common structural framework around which diversity is built via different building blocks") h1(doc, "3. TYPES OF COMBINATORIAL LIBRARIES") h2(doc, "3.1 Peptide Libraries") body(doc, "First combinatorial libraries; systematic variation of amino acids in a peptide chain") code_block(doc, "Dipeptide library using 20 amino acids:\n20 × 20 = 400 dipeptides\n\nTripeptide library:\n20 × 20 × 20 = 8,000 tripeptides\n\nHexapeptide library:\n20⁶ = 64,000,000 hexapeptides\n\nPin method (Geysen, 1984):\n 96-well polypropylene pins\n Each pin: independent solid-phase synthesis site\n 96 peptides synthesised simultaneously") h2(doc, "3.2 Small Molecule Libraries") code_block(doc, "Benzodiazepine library (Bunin & Ellman, 1992):\n\n Step 1: Couple amino acid to resin\n Step 2: Cyclise with nitrobenzodiazepine\n Step 3: Acylate N-1\n → 192 benzodiazepines from 7-step solid-phase synthesis\n\nOther small molecule scaffolds used:\n• Dihydropyridines\n• Pyrrolidines\n• Piperazines\n• Indoles\n• Purines\n• Thiazolidines") h2(doc, "3.3 Oligonucleotide Libraries") body(doc, "DNA/RNA oligomers; used in SELEX (systematic evolution of ligands by exponential enrichment)") h1(doc, "4. METHODS OF COMBINATORIAL SYNTHESIS") h2(doc, "4.1 Solid-Phase Synthesis (SPS)") body(doc, "Compounds are synthesised while attached to insoluble polymer support (resin). Excess reagents washed away; product remains on resin until cleavage.") code_block(doc, "Solid-Phase Synthesis Steps:\n\n Resin─Linker─[Functional group]\n ↓ Attach first building block (BB₁)\n Resin─Linker─BB₁─[Protected]\n ↓ Deprotect\n Resin─Linker─BB₁─[Free NH/OH]\n ↓ Couple BB₂\n Resin─Linker─BB₁─BB₂\n ↓ Repeat n times\n Resin─Linker─BB₁─BB₂─...─BBn\n ↓ Cleave from resin (TFA, HF, photolysis)\n Free compound: BB₁─BB₂─...─BBn\n\nResins used:\n• Wang resin (acid-cleavable with TFA)\n• Rink amide resin (gives C-terminal amide)\n• Merrifield resin (HF cleavage)\n• TentaGel resin (PEG-PS, for aqueous reactions)\n• Rasta resin, ArgoPore") h2(doc, "4.2 Solution-Phase Synthesis") body(doc, "Reactions carried out in solution (traditional flasks); products purified after each step") bullet(doc, "Advantage: All reaction types compatible; easier scale-up") bullet(doc, "Disadvantage: Time-consuming purification; not as parallel") bullet(doc, "Polymer-assisted solution synthesis (PASS): uses scavenger resins to capture excess reagents") h2(doc, "4.3 Split-and-Pool (Mix-and-Split) Method") body(doc, "Most powerful method; generates enormous libraries from few building blocks") code_block(doc, "Split-and-Pool Protocol (Furka, 1991):\n\n Step 1: SPLIT resin beads into n equal portions (n = no. of BBs)\n\n ┌─────────┐ ┌─────────┐ ┌─────────┐\n │ Portion1│ │ Portion2│ │ Portion3│\n └─────────┘ └─────────┘ └─────────┘\n\n Step 2: React each portion with different building block\n ↓BB₁ ↓BB₂ ↓BB₃\n Resin-BB₁ Resin-BB₂ Resin-BB₃\n\n Step 3: POOL (mix) all portions together\n Resin mixture: (BB₁ + BB₂ + BB₃) beads\n\n Step 4: Re-SPLIT and react with next set of BBs\n ↓BB₄ ↓BB₅ ↓BB₆\n\n Continue for desired library size\n\n Final library size = n₁ × n₂ × n₃ × ... × nₓ\n e.g. 3 × 3 × 3 = 27 compounds from 9 building blocks") bullet(doc, "One bead – one compound (OBOC): each bead carries single compound") bullet(doc, "Deconvolution needed to identify active compounds") bullet(doc, "Encoded libraries: tag each bead with identifier (chemical, radio-frequency)") h2(doc, "4.4 Parallel Synthesis") body(doc, "Each compound synthesised in a separate vessel simultaneously; one compound per vessel") code_block(doc, "Parallel synthesis formats:\n\n• Multi-well plates (96, 384 well plates)\n• Reaction blocks (24, 48, 96 vessels)\n• Automated synthesisers (IRORI, Tecan, Bohdan)\n• Flow chemistry reactors\n\nAdvantage: Each compound is pure (no deconvolution)\nDisadvantage: Smaller library size vs split-and-pool\n\nTypical format: 96-well plate → 96 compounds per run\n Combined over multiple runs → thousands of compounds") h1(doc, "5. SOLID-PHASE REAGENTS AND LINKERS") h2(doc, "5.1 Linker Types") add_table(doc, ["Linker", "Cleavage Condition", "Released Group"], [ ["Wang linker", "TFA / DCM", "Carboxylic acid"], ["Rink amide linker", "TFA / DCM", "Primary amide"], ["PAM linker", "HF or TFA", "Carboxylic acid"], ["Photolabile linker", "hν (UV light)", "Acid / alcohol"], ["Safety-catch linker", "Activation then nucleophile", "Amide/ester"], ["Hydrazide linker", "Acid/aldehyde", "Hydrazide"], ["Traceless linker", "Various", "C–H bond (no tag)"], ] ) h2(doc, "5.2 Scavenger Resins") body(doc, "Used to remove excess reagents from solution-phase reactions") code_block(doc, "Examples:\n• Amine scavenger: aldehyde resin (removes excess amines)\n• Isocyanate resin (removes amines)\n• Carbonate resin (removes alcohols)\n• MP-carbonate (removes acids)") h1(doc, "6. HIGH-THROUGHPUT SCREENING (HTS)") body(doc, "Rapid automated testing of large compound libraries against a biological target") code_block(doc, "HTS Workflow:\n\n Combinatorial library\n ↓\n Miniaturised assay (384 / 1536 well plates)\n ↓\n Robotic liquid handling + automated reader\n ↓\n Primary screen: single conc. → % inhibition\n ↓\n Hit selection (e.g. >50% inhibition)\n ↓\n Confirmation screen (duplicate, multiple conc.)\n ↓\n Dose-response curves → IC₅₀, Ki\n ↓\n Hit-to-Lead optimisation") bullet(doc, "Z-factor (Z'): statistical measure of HTS quality") code_block(doc, "Z' = 1 – 3(σ_pos + σ_neg) / |μ_pos – μ_neg|\nZ' > 0.5 → excellent assay; Z' = 1 → perfect assay") bullet(doc, "Ultra-HTS: >100,000 compounds/day; 1536-well format") h1(doc, "7. ENCODING STRATEGIES") body(doc, "Methods to identify which compound on a bead is active after biological screening") h2(doc, "7.1 Chemical Encoding") bullet(doc, "Attach chemical 'tags' (haloaromatic amides, peptide tags) to bead during synthesis") bullet(doc, "Decode by GC-MS or LC-MS after hit identification") h2(doc, "7.2 Oligonucleotide (DNA) Encoding") bullet(doc, "Attach DNA 'barcode' sequence to each bead") bullet(doc, "PCR amplification + sequencing to decode") h2(doc, "7.3 Radio-Frequency (RF) Encoding") bullet(doc, "IRORI technology: microreactors with RF transponder chips") bullet(doc, "RF tag records synthesis history of each microreactor") bullet(doc, "Decode by RF reader after screening") h2(doc, "7.4 Positional/Spatial Encoding") bullet(doc, "Location in plate encodes identity (parallel synthesis)") bullet(doc, "No decoding needed – position = structure") h1(doc, "8. DECONVOLUTION STRATEGIES") body(doc, "Methods to identify active compound within a mixture (pool/library)") h2(doc, "8.1 Iterative Deconvolution") code_doc, "Prepare mixtures with one position held constant at each iteration:\n\nLibrary: 3 positions, 3 BBs each = 27 compounds\nRound 1: fix position 1 (3 sub-libraries of 9 compounds each)\n → identify most active subgroup\nRound 2: fix position 1 + 2 (3 sub-libraries of 3 compounds)\n → identify next best\nRound 3: fix all positions → single compound identified" # Use body since code block needs proper variable name fix body(doc, "Iterative deconvolution: prepare sub-libraries fixing one position at a time. Round 1 fixes position 1 → 3 pools of 9; most active pool selected. Round 2 fixes positions 1+2 → 3 pools of 3; most active selected. Round 3 fixes all → single compound. Each round synthesise fresh sub-libraries and test.") h2(doc, "8.2 Positional Scanning") bullet(doc, "All positions scanned simultaneously (not sequentially)") bullet(doc, "Faster but requires more synthesis") h2(doc, "8.3 Recursive Deconvolution") bullet(doc, "Uses activity data recursively to narrow down active compound") bullet(doc, "Mathematical approach; no re-synthesis needed") h1(doc, "9. DIVERSITY-ORIENTED SYNTHESIS (DOS)") body(doc, "Aims to produce structurally diverse small-molecule libraries that explore broad regions of chemical space") code_block(doc, "DOS features:\n• Scaffold diversity (different ring systems)\n• Stereochemical diversity\n• Appendage diversity (different substituents)\n\nCompare:\nTarget-Oriented Synthesis (TOS): one specific target\nDiversity-Oriented Synthesis (DOS): broad coverage of chemical space\n\nBranching point strategy: one scaffold → multiple ring systems via\n • Cyclisation conditions\n • Ring-closing metathesis (Grubbs catalyst)\n • Diels-Alder reaction\n • Intramolecular Mannich reaction") h1(doc, "10. FRAGMENT-BASED DRUG DISCOVERY (FBDD)") body(doc, "Uses very small molecules (fragments, MW ~100-300 Da) as starting points; weak binders (mM Kd) identified and grown/linked into potent drugs") code_block(doc, "FBDD vs HTS:\n FBDD HTS\nLibrary size: ~1,000 ~100,000s\nMW (Da): 100–300 300–600\nBinding affinity: mM μM–nM\nHit rate: 1–20% 0.01–0.1%\nLigand Efficiency: High Lower\n\nLigand Efficiency (LE) = ΔG_binding / heavy atom count\n\nFragment detection methods:\n• SPR (Surface Plasmon Resonance)\n• NMR (STD NMR, WaterLOGSY, ¹⁵N HSQC)\n• X-ray crystallography\n• DSF (Differential Scanning Fluorimetry)\n• MST (Micro-Scale Thermophoresis)") h1(doc, "11. APPLICATIONS") bullet(doc, "Lead compound identification (HTS + CombiChem)") bullet(doc, "Optimisation of ADMET properties") bullet(doc, "Agrochemicals and material science") bullet(doc, "Peptide/peptidomimetic libraries") bullet(doc, "Catalyst screening (combinatorial catalysis)") bullet(doc, "Antibody library generation (phage display)") h1(doc, "12. ADVANTAGES AND LIMITATIONS") h2(doc, "Advantages") bullet(doc, "Rapid generation of large compound collections") bullet(doc, "Increased speed of lead identification") bullet(doc, "Cost-effective per compound synthesised") bullet(doc, "Amenable to automation and robotics") bullet(doc, "Can explore vast chemical space") h2(doc, "Limitations") bullet(doc, "Libraries may have poor drug-likeness (need Ro5 filtering)") bullet(doc, "Purity issues in pool/mixture libraries") bullet(doc, "Deconvolution is time-consuming and complex") bullet(doc, "Redundancy in library design") bullet(doc, "Bias towards easily accessible chemistry") bullet(doc, "False positives in HTS (aggregators, fluorescent compounds, reactive compounds)") doc.save("/tmp/workspace/notes_project/4_CombChem_Notes.docx") print("CombChem done") print("\nAll 4 documents saved to /tmp/workspace/notes_project/") PYEOF

python3 << 'PYEOF' from docx import Document from docx.shared import Pt, Inches, RGBColor from docx.enum.text import WD_ALIGN_PARAGRAPH import os def new_doc(title): doc = Document() ns = doc.styles["Normal"] ns.font.name = "Arial" ns.font.size = Pt(11) h = doc.add_heading(title, level=0) h.alignment = WD_ALIGN_PARAGRAPH.CENTER for run in h.runs: run.font.color.rgb = RGBColor(0x1F, 0x49, 0x7D) doc.add_paragraph() return doc def h1(doc, text): p = doc.add_heading(text, level=1) for run in p.runs: run.font.color.rgb = RGBColor(0x1F, 0x49, 0x7D) return p def h2(doc, text): p = doc.add_heading(text, level=2) for run in p.runs: run.font.color.rgb = RGBColor(0xC5, 0x50, 0x00) return p def h3(doc, text): return doc.add_heading(text, level=3) def body(doc, text): return doc.add_paragraph(text) def bullet(doc, text, level=0): p = doc.add_paragraph(text, style="List Bullet") p.paragraph_format.left_indent = Inches(0.25 * (level + 1)) return p def code_block(doc, text): p = doc.add_paragraph() run = p.add_run(text) run.font.name = "Courier New" run.font.size = Pt(9) p.paragraph_format.left_indent = Inches(0.4) from docx.oxml.ns import qn from docx.oxml import OxmlElement pPr = p._p.get_or_add_pPr() shd = OxmlElement("w:shd") shd.set(qn("w:val"), "clear") shd.set(qn("w:color"), "auto") shd.set(qn("w:fill"), "F2F2F2") pPr.append(shd) return p def add_table(doc, headers, rows): table = doc.add_table(rows=1+len(rows), cols=len(headers)) table.style = "Light Shading Accent 1" hdr = table.rows[0].cells for i, h in enumerate(headers): hdr[i].text = h for run in hdr[i].paragraphs[0].runs: run.font.bold = True for ri, row in enumerate(rows): cells = table.rows[ri+1].cells for ci, val in enumerate(row): cells[ci].text = val doc.add_paragraph() # ================================================================= # DOCUMENT 4 – COMBINATORIAL CHEMISTRY # ================================================================= doc = new_doc("COMBINATORIAL CHEMISTRY\nDeep-Dive Study Notes") h1(doc, "1. INTRODUCTION") body(doc, "Combinatorial chemistry is a technique by which a large number of structurally diverse small molecules (libraries) can be synthesised simultaneously using systematic and repetitive chemical synthesis. It accelerates drug discovery by rapidly generating large compound collections for biological screening.") bullet(doc, "Traditional synthesis: 1 compound at a time") bullet(doc, "Combinatorial synthesis: hundreds to millions of compounds simultaneously") bullet(doc, "Combined with High-Throughput Screening (HTS) → rapid hit identification") bullet(doc, "Emerged in 1990s; revolutionised pharmaceutical lead identification") bullet(doc, "Pioneers: R.B. Merrifield (solid-phase peptide synthesis, Nobel 1984), Geysen, Furka") h1(doc, "2. KEY CONCEPTS") h2(doc, "2.1 Library") body(doc, "A collection of compounds prepared by combinatorial methods, all sharing a common scaffold but varying substituents") code_block(doc, "Scaffold (Core) + Varying Building Blocks → Library\n\nExample: Benzodiazepine library\n\n R₁\n |\n N─C\n / \\\n C C=O\n | |\n C NH\n \\ /\n C=C\n |\n R₂\n\nVary R₁ and R₂ → large benzodiazepine library") h2(doc, "2.2 Building Blocks (Monomers)") body(doc, "Reactant molecules combined in different combinations to generate library members. Must be commercially available, reactive, and structurally diverse.") h2(doc, "2.3 Scaffold / Core Structure") body(doc, "The common structural framework around which diversity is built via different building blocks") h1(doc, "3. TYPES OF COMBINATORIAL LIBRARIES") h2(doc, "3.1 Peptide Libraries") body(doc, "First combinatorial libraries; systematic variation of amino acids in a peptide chain") code_block(doc, "Dipeptide library using 20 amino acids:\n20 x 20 = 400 dipeptides\n\nTripeptide library:\n20 x 20 x 20 = 8,000 tripeptides\n\nHexapeptide library:\n20^6 = 64,000,000 hexapeptides\n\nPin method (Geysen, 1984):\n 96-well polypropylene pins\n Each pin: independent solid-phase synthesis site\n 96 peptides synthesised simultaneously") h2(doc, "3.2 Small Molecule Libraries") code_block(doc, "Benzodiazepine library (Bunin & Ellman, 1992):\n\n Step 1: Couple amino acid to resin\n Step 2: Cyclise with nitrobenzodiazepine\n Step 3: Acylate N-1\n --> 192 benzodiazepines from 7-step solid-phase synthesis\n\nOther small molecule scaffolds:\n Dihydropyridines, Pyrrolidines, Piperazines,\n Indoles, Purines, Thiazolidines") h2(doc, "3.3 Oligonucleotide Libraries") body(doc, "DNA/RNA oligomers; used in SELEX (Systematic Evolution of Ligands by Exponential enrichment)") h1(doc, "4. METHODS OF COMBINATORIAL SYNTHESIS") h2(doc, "4.1 Solid-Phase Synthesis (SPS)") body(doc, "Compounds are synthesised while attached to insoluble polymer support (resin). Excess reagents washed away; product remains on resin until cleavage.") code_block(doc, "Solid-Phase Synthesis Steps:\n\n Resin-Linker-[Functional group]\n | Attach first building block (BB1)\n Resin-Linker-BB1-[Protected]\n | Deprotect\n Resin-Linker-BB1-[Free NH/OH]\n | Couple BB2\n Resin-Linker-BB1-BB2\n | Repeat n times\n Resin-Linker-BB1-BB2-...-BBn\n | Cleave from resin (TFA, HF, photolysis)\n Free compound: BB1-BB2-...-BBn\n\nResins used:\n Wang resin (acid-cleavable, TFA)\n Rink amide resin (gives C-terminal amide)\n Merrifield resin (HF cleavage)\n TentaGel resin (PEG-PS, aqueous reactions)") h2(doc, "4.2 Solution-Phase Synthesis") body(doc, "Reactions carried out in solution (traditional flasks); products purified after each step") bullet(doc, "Advantage: All reaction types compatible; easier scale-up") bullet(doc, "Disadvantage: Time-consuming purification; not as parallel") bullet(doc, "Polymer-assisted solution synthesis (PASS): uses scavenger resins") h2(doc, "4.3 Split-and-Pool (Mix-and-Split) Method") body(doc, "Most powerful method; generates enormous libraries from few building blocks (Furka, 1991)") code_block(doc, "Split-and-Pool Protocol:\n\n Step 1: SPLIT resin beads into n equal portions\n\n [Portion 1] [Portion 2] [Portion 3]\n\n Step 2: React each portion with different BB\n +BB1 +BB2 +BB3\n Resin-BB1 Resin-BB2 Resin-BB3\n\n Step 3: POOL (mix) all portions together\n [BB1+BB2+BB3 bead mixture]\n\n Step 4: Re-SPLIT and react with next BBs\n +BB4 +BB5 +BB6\n\n Continue for desired library depth\n\n Final library size = n1 x n2 x n3 x ... x nx\n Example: 3 x 3 x 3 = 27 compounds from 9 BBs\n\nOne bead - one compound (OBOC) principle") bullet(doc, "Deconvolution needed to identify active compound in pool") bullet(doc, "Encoded libraries: tag each bead (chemical tag, RF tag, DNA barcode)") h2(doc, "4.4 Parallel Synthesis") body(doc, "Each compound synthesised in a separate vessel simultaneously; one compound per vessel") code_block(doc, "Parallel synthesis formats:\n\n 96-well plates → 96 compounds per run\n 384-well plates → 384 compounds per run\n Automated synthesisers: IRORI, Tecan, Bohdan\n Flow chemistry reactors\n\nAdvantage: Each compound is pure (no deconvolution needed)\nDisadvantage: Smaller library vs split-and-pool") h1(doc, "5. LINKERS IN SOLID-PHASE SYNTHESIS") add_table(doc, ["Linker", "Cleavage Condition", "Released Group"], [ ["Wang linker", "TFA / DCM", "Carboxylic acid"], ["Rink amide linker", "TFA / DCM", "Primary amide"], ["PAM linker", "HF or TFA", "Carboxylic acid"], ["Photolabile linker", "UV light (hv)", "Acid / alcohol"], ["Safety-catch linker", "Activation + nucleophile", "Amide / ester"], ["Traceless linker", "Various", "C-H bond (no tag left)"], ] ) h1(doc, "6. HIGH-THROUGHPUT SCREENING (HTS)") body(doc, "Rapid automated testing of large compound libraries against biological targets") code_block(doc, "HTS Workflow:\n\n Combinatorial library\n |\n Miniaturised assay (384 / 1536 well plates)\n |\n Robotic liquid handling + automated plate reader\n |\n Primary screen: single concentration -> % inhibition\n |\n Hit selection (threshold: e.g. >50% inhibition)\n |\n Confirmation screen (duplicate, dose-response)\n |\n IC50, Ki determination\n |\n Hit-to-Lead optimisation") code_block(doc, "Z-factor (Z') = assay quality metric:\n\nZ' = 1 - [3(sigma_pos + sigma_neg)] / |mu_pos - mu_neg|\n\nwhere sigma = std deviation, mu = mean of positive/negative controls\n\nZ' > 0.5 --> Excellent assay\nZ' = 1.0 --> Perfect assay\nZ' < 0.5 --> Poor assay (needs optimisation)") h1(doc, "7. ENCODING AND DECONVOLUTION STRATEGIES") h2(doc, "7.1 Encoding Methods") add_table(doc, ["Method", "Tag Used", "Decoding"], [ ["Chemical encoding", "Haloaromatic amide tags", "GC-MS / LC-MS"], ["DNA encoding", "Oligonucleotide barcode", "PCR + DNA sequencing"], ["RF encoding (IRORI)", "Radio-frequency transponder chip", "RF reader"], ["Positional encoding", "Location in plate (parallel)", "Position = structure"], ] ) h2(doc, "7.2 Deconvolution Strategies") body(doc, "Methods to identify which compound in a pool is responsible for activity:") bullet(doc, "Iterative deconvolution: fix one position at a time per round; synthesise fresh sub-libraries; test; narrow down to single compound") bullet(doc, "Positional scanning: all positions scanned simultaneously (more synthesis but faster)") bullet(doc, "Recursive deconvolution: mathematical analysis of activity data without re-synthesis") bullet(doc, "OBOC + affinity selection: active bead isolated by binding to labelled target, bead decoded by MS/sequencing") h1(doc, "8. DIVERSITY-ORIENTED SYNTHESIS (DOS)") body(doc, "Aims to produce structurally diverse libraries exploring broad chemical space – contrast with Target-Oriented Synthesis (TOS)") code_block(doc, "DOS features:\n • Scaffold diversity (different ring systems)\n • Stereochemical diversity (multiple stereocentres)\n • Appendage diversity (different substituents)\n\nBranching point strategy:\n One scaffold --> multiple ring systems via:\n - Ring-closing metathesis (Grubbs catalyst)\n - Diels-Alder cycloaddition\n - Intramolecular Mannich reaction\n - Reductive amination\n\nTOS: one specific target molecule (natural product synthesis)\nDOS: broad chemical space coverage (lead identification)") h1(doc, "9. FRAGMENT-BASED DRUG DISCOVERY (FBDD)") body(doc, "Uses very small molecules (fragments, MW 100-300 Da) as starting points; weak binders (mM Kd) identified and grown/linked into potent drugs") code_block(doc, " FBDD HTS\n---------------------------------------------\nLibrary size: ~1,000 ~100,000+\nMW (Da): 100-300 300-600\nBinding affinity: mM uM-nM\nHit rate: 1-20% 0.01-0.1%\nLigand Efficiency: High Lower\n\nLigand Efficiency (LE) = DeltaG_binding / heavy atom count\n\nFragment detection methods:\n SPR (Surface Plasmon Resonance)\n NMR (STD-NMR, WaterLOGSY, 15N-HSQC)\n X-ray crystallography\n DSF (Differential Scanning Fluorimetry)\n MST (Micro-Scale Thermophoresis)\n\nFBDD success: Vemurafenib (B-RAF inhibitor)") h1(doc, "10. APPLICATIONS OF COMBINATORIAL CHEMISTRY") bullet(doc, "Lead compound identification (HTS + combinatorial chemistry)") bullet(doc, "Optimisation of ADMET properties (SAR library around lead)") bullet(doc, "Agrochemicals and pesticide libraries") bullet(doc, "Peptide and peptidomimetic libraries") bullet(doc, "Combinatorial catalysis (catalyst screening)") bullet(doc, "Antibody library generation (phage display)") bullet(doc, "Materials science (combinatorial material discovery)") h1(doc, "11. ADVANTAGES AND LIMITATIONS") h2(doc, "Advantages") bullet(doc, "Rapid generation of large, diverse compound collections") bullet(doc, "Increased speed of lead identification vs traditional medicinal chemistry") bullet(doc, "Cost-effective per compound synthesised") bullet(doc, "Fully amenable to automation and robotics") bullet(doc, "Enables exploration of vast regions of chemical space") h2(doc, "Limitations") bullet(doc, "Libraries may lack drug-likeness (Ro5 filtering required)") bullet(doc, "Purity issues in pool/mixture libraries") bullet(doc, "Deconvolution is time-consuming and complex") bullet(doc, "Redundancy in library design reduces efficiency") bullet(doc, "Bias towards easily accessible chemistry / reactions") bullet(doc, "False positives in HTS (aggregators, fluorescent compounds, reactive compounds)") bullet(doc, "PAINS (Pan-Assay Interference Compounds) are common in combinatorial libraries") doc.save("/tmp/workspace/notes_project/4_CombChem_Notes.docx") print("CombChem done") # Verify all files for fn in ["1_QSAR_Notes.docx", "2_ProDrug_Notes.docx", "3_CADD_Notes.docx", "4_CombChem_Notes.docx"]: path = f"/tmp/workspace/notes_project/{fn}" size = os.path.getsize(path) print(f"{fn}: {size} bytes") PYEOF