Update app.py

app.py

@@ -9,7 +9,12 @@ from io import BytesIO
 from PIL import Image
 from datetime import datetime
 from pathlib import Path
+import os
 
+# Вимкнення телеметрії Gradio (щоб уникнути непотрібних запитів)
+os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
+
+# Кольори (ваші)
 BG   = "#0f172a"
 CARD = "#1e293b"
 ACC  = "#f97316"
@@ -22,6 +27,7 @@ BORDER = "#334155"
 
 LOG_PATH = Path("/tmp/lab_journal.csv")
 
+# ---------- Логування ----------
 def log_entry(tab, inputs, result, note=""):
     try:
         write_header = not LOG_PATH.exists()
@@ -45,7 +51,7 @@ def save_note(note, tab, last_result):
     log_entry(tab, "", last_result, note)
     return "✅ Saved!", load_journal()
 
-# ── DATABASES ─────────────────────────────────────────────────────────────────
+# ---------- Бази даних (ваші) ----------
 MIRNA_DB = {
     "BRCA2": [
         {"miRNA":"hsa-miR-148a-3p","log2FC":-0.70,"padj":0.013,"targets":"DNMT1, AKT2","pathway":"Epigenetic reprogramming"},
@@ -69,6 +75,7 @@ MIRNA_DB = {
         {"miRNA":"hsa-miR-215-5p","log2FC":-0.51,"padj":0.038,"targets":"DTL, DHFR","pathway":"DNA damage response"},
     ],
 }
+
 SIRNA_DB = {
     "LUAD": [
         {"Gene":"SPC24","dCERES":-0.175,"log2FC":1.13,"Drug_status":"Novel","siRNA":"GCAGCUGAAGAAACUGAAU"},
@@ -92,6 +99,7 @@ SIRNA_DB = {
         {"Gene":"PKMYT1","dCERES":-0.122,"log2FC":1.07,"Drug_status":"Clinical","siRNA":"GACGCUCAAGAUGCAGAUU"},
     ],
 }
+
 CERNA = [
     {"lncRNA":"CYTOR","miRNA":"hsa-miR-138-5p","target":"AKT1","pathway":"TREM2 core signaling"},
     {"lncRNA":"CYTOR","miRNA":"hsa-miR-138-5p","target":"NFKB1","pathway":"Neuroinflammation"},
@@ -99,6 +107,7 @@ CERNA = [
     {"lncRNA":"GAS5","miRNA":"hsa-miR-222-3p","target":"IL1B","pathway":"Neuroinflammation"},
     {"lncRNA":"HOTAIRM1","miRNA":"hsa-miR-9-5p","target":"TREM2","pathway":"Direct TREM2 regulation"},
 ]
+
 ASO = [
     {"lncRNA":"GAS5","position":119,"accessibility":0.653,"GC_pct":50,"Tm":47.2,"priority":"HIGH"},
     {"lncRNA":"CYTOR","position":507,"accessibility":0.653,"GC_pct":50,"Tm":46.8,"priority":"HIGH"},
@@ -106,6 +115,7 @@ ASO = [
     {"lncRNA":"LINC00847","position":89,"accessibility":0.598,"GC_pct":56,"Tm":48.3,"priority":"MEDIUM"},
     {"lncRNA":"ZFAS1","position":312,"accessibility":0.571,"GC_pct":48,"Tm":45.5,"priority":"MEDIUM"},
 ]
+
 FGFR3 = {
     "P1 (hairpin loop)": [
         {"Compound":"CHEMBL1575701","RNA_score":0.809,"Toxicity":0.01,"Final_score":0.793},
@@ -122,6 +132,7 @@ FGFR3 = {
         {"Compound":"Berberine","RNA_score":0.735,"Toxicity":3.2,"Final_score":0.708},
     ],
 }
+
 VARIANT_DB = {
     "BRCA1:p.R1699Q": {"score":0.03,"cls":"Benign","conf":"High"},
     "BRCA1:p.R1699W": {"score":0.97,"cls":"Pathogenic","conf":"High"},
@@ -131,22 +142,71 @@ VARIANT_DB = {
     "EGFR:p.L858R":   {"score":0.96,"cls":"Pathogenic","conf":"High"},
     "ALK:p.F1174L":   {"score":0.94,"cls":"Pathogenic","conf":"High"},
 }
+
 PLAIN = {
     "Pathogenic":        "This variant is likely to cause disease. Clinical follow-up is strongly recommended.",
     "Likely Pathogenic": "This variant is probably harmful. Discuss with your doctor.",
     "Benign":            "This variant is likely harmless. Common in the general population.",
     "Likely Benign":     "This variant is probably harmless. No strong reason for concern.",
 }
+
 BM_W = {
     "CTHRC1":0.18,"FHL2":0.15,"LDHA":0.14,"P4HA1":0.13,
     "SERPINH1":0.12,"ABCA8":-0.11,"CA4":-0.10,"CKB":-0.09,
     "NNMT":0.08,"CACNA2D2":-0.07
 }
+
 PROTEINS = ["albumin","apolipoprotein","fibrinogen","vitronectin",
             "clusterin","igm","iga","igg","complement","transferrin",
             "alpha-2-macroglobulin"]
 
-# ── LOGIC ─────────────────────────────────────────────────────────────────────
+# ---------- Дані для рідкісних раків ----------
+DIPG_VARIANTS = [
+    {"Variant":"H3K27M (H3F3A)","Freq_pct":78,"Pathway":"PRC2 inhibition → global H3K27me3 loss","Drug_status":"ONC201 (clinical)","Circadian_gene":"BMAL1 suppressed"},
+    {"Variant":"ACVR1 p.R206H","Freq_pct":21,"Pathway":"BMP/SMAD hyperactivation","Drug_status":"LDN-193189 (preclinical)","Circadian_gene":"PER1 disrupted"},
+    {"Variant":"PIK3CA p.H1047R","Freq_pct":15,"Pathway":"PI3K/AKT/mTOR","Drug_status":"Copanlisib (clinical)","Circadian_gene":"CRY1 altered"},
+    {"Variant":"TP53 p.R248W","Freq_pct":14,"Pathway":"DNA damage response loss","Drug_status":"APR-246 (clinical)","Circadian_gene":"p53-CLOCK axis"},
+    {"Variant":"PDGFRA amp","Freq_pct":13,"Pathway":"RTK/RAS signalling","Drug_status":"Avapritinib (clinical)","Circadian_gene":"REV-ERB altered"},
+]
+
+DIPG_CSF_LNP = [
+    {"Formulation":"MC3-DSPC-Chol-PEG","Size_nm":92,"Zeta_mV":-4.1,"CSF_protein":"Beta2-microglobulin","ApoE_pct":12.4,"BBB_est":0.41,"Priority":"HIGH"},
+    {"Formulation":"DLin-KC2-DSPE-PEG","Size_nm":87,"Zeta_mV":-3.8,"CSF_protein":"Cystatin C","ApoE_pct":14.1,"BBB_est":0.47,"Priority":"HIGH"},
+    {"Formulation":"C12-200-DOPE-PEG","Size_nm":103,"Zeta_mV":-5.2,"CSF_protein":"Albumin (low)","ApoE_pct":9.8,"BBB_est":0.33,"Priority":"MEDIUM"},
+    {"Formulation":"DODAP-DSPC-Chol","Size_nm":118,"Zeta_mV":-2.1,"CSF_protein":"Transferrin","ApoE_pct":7.2,"BBB_est":0.24,"Priority":"LOW"},
+]
+
+UVM_VARIANTS = [
+    {"Variant":"GNAQ p.Q209L","Freq_pct":46,"Pathway":"PLCβ → PKC → MAPK","Drug_status":"Darovasertib (clinical)","m6A_writer":"METTL3 upregulated"},
+    {"Variant":"GNA11 p.Q209L","Freq_pct":32,"Pathway":"PLCβ → PKC → MAPK","Drug_status":"Darovasertib (clinical)","m6A_writer":"WTAP upregulated"},
+    {"Variant":"BAP1 loss","Freq_pct":47,"Pathway":"Chromatin remodeling → metastasis","Drug_status":"No approved (HDAC trials)","m6A_writer":"FTO overexpressed"},
+    {"Variant":"SF3B1 p.R625H","Freq_pct":19,"Pathway":"Splicing alteration → neoepitopes","Drug_status":"H3B-8800 (clinical)","m6A_writer":"METTL14 altered"},
+    {"Variant":"EIF1AX p.A113_splice","Freq_pct":14,"Pathway":"Translation initiation","Drug_status":"Novel — no drug","m6A_writer":"YTHDF2 suppressed"},
+]
+
+UVM_VITREOUS_LNP = [
+    {"Formulation":"SM-102-DSPC-Chol-PEG","Vitreal_protein":"Hyaluronan-binding","Size_nm":95,"Zeta_mV":-3.2,"Retention_h":18,"Priority":"HIGH"},
+    {"Formulation":"Lipid-H-DOPE-PEG","Vitreal_protein":"Vitronectin dominant","Size_nm":88,"Zeta_mV":-4.0,"Retention_h":22,"Priority":"HIGH"},
+    {"Formulation":"DOTAP-DSPC-PEG","Vitreal_protein":"Albumin wash-out","Size_nm":112,"Zeta_mV":+2.1,"Retention_h":6,"Priority":"LOW"},
+    {"Formulation":"MC3-DPPC-Chol","Vitreal_protein":"Clusterin-rich","Size_nm":101,"Zeta_mV":-2.8,"Retention_h":14,"Priority":"MEDIUM"},
+]
+
+PAML_VARIANTS = [
+    {"Variant":"FLT3-ITD","Freq_pct":25,"Pathway":"RTK constitutive activation → JAK/STAT","Drug_status":"Midostaurin (approved)","Ferroptosis":"GPX4 suppressed"},
+    {"Variant":"NPM1 c.860_863dupTCAG","Freq_pct":30,"Pathway":"Nuclear export deregulation","Drug_status":"APR-548 combo (clinical)","Ferroptosis":"SLC7A11 upregulated"},
+    {"Variant":"DNMT3A p.R882H","Freq_pct":18,"Pathway":"Epigenetic dysregulation","Drug_status":"Azacitidine (approved)","Ferroptosis":"ACSL4 altered"},
+    {"Variant":"CEBPA biallelic","Freq_pct":8,"Pathway":"Myeloid differentiation block","Drug_status":"Novel target","Ferroptosis":"NRF2 pathway"},
+    {"Variant":"IDH1/2 mutation","Freq_pct":15,"Pathway":"2-HG oncometabolite → TET2 inhibition","Drug_status":"Enasidenib (approved)","Ferroptosis":"Iron metabolism disrupted"},
+]
+
+PAML_BM_LNP = [
+    {"Formulation":"ALC-0315-DSPC-Chol-PEG","BM_protein":"ApoE + Clusterin","Size_nm":98,"Zeta_mV":-3.5,"Marrow_uptake_pct":34,"Priority":"HIGH"},
+    {"Formulation":"MC3-DOPE-Chol-PEG","BM_protein":"Fibronectin dominant","Size_nm":105,"Zeta_mV":-4.2,"Marrow_uptake_pct":28,"Priority":"HIGH"},
+    {"Formulation":"DLin-MC3-DPPC","BM_protein":"Vitronectin-rich","Size_nm":91,"Zeta_mV":-2.9,"Marrow_uptake_pct":19,"Priority":"MEDIUM"},
+    {"Formulation":"Cationic-DOTAP-Chol","BM_protein":"Opsonin-heavy","Size_nm":132,"Zeta_mV":+8.1,"Marrow_uptake_pct":8,"Priority":"LOW"},
+]
+
+# ---------- Функції логіки (ваші, без змін) ----------
 def predict_mirna(gene):
     df = pd.DataFrame(MIRNA_DB.get(gene, []))
     log_entry("S1-B | S1-R2 | miRNA", gene, f"{len(df)} miRNAs")
@@ -300,55 +360,7 @@ def extract_corona(text):
     log_entry("S1-D | S1-R10 | NLP", text[:80], f"proteins={len(out['corona_proteins'])}")
     return json.dumps(out, indent=2), summary
 
-
-# ── S1-F · PHYLO-RARE DATABASES ───────────────────────────────────────────────
-
-# S1-R12b · DIPG — H3K27M + CSF LNP + Circadian
-DIPG_VARIANTS = [
-    {"Variant":"H3K27M (H3F3A)","Freq_pct":78,"Pathway":"PRC2 inhibition → global H3K27me3 loss","Drug_status":"ONC201 (clinical)","Circadian_gene":"BMAL1 suppressed"},
-    {"Variant":"ACVR1 p.R206H","Freq_pct":21,"Pathway":"BMP/SMAD hyperactivation","Drug_status":"LDN-193189 (preclinical)","Circadian_gene":"PER1 disrupted"},
-    {"Variant":"PIK3CA p.H1047R","Freq_pct":15,"Pathway":"PI3K/AKT/mTOR","Drug_status":"Copanlisib (clinical)","Circadian_gene":"CRY1 altered"},
-    {"Variant":"TP53 p.R248W","Freq_pct":14,"Pathway":"DNA damage response loss","Drug_status":"APR-246 (clinical)","Circadian_gene":"p53-CLOCK axis"},
-    {"Variant":"PDGFRA amp","Freq_pct":13,"Pathway":"RTK/RAS signalling","Drug_status":"Avapritinib (clinical)","Circadian_gene":"REV-ERB altered"},
-]
-DIPG_CSF_LNP = [
-    {"Formulation":"MC3-DSPC-Chol-PEG","Size_nm":92,"Zeta_mV":-4.1,"CSF_protein":"Beta2-microglobulin","ApoE_pct":12.4,"BBB_est":0.41,"Priority":"HIGH"},
-    {"Formulation":"DLin-KC2-DSPE-PEG","Size_nm":87,"Zeta_mV":-3.8,"CSF_protein":"Cystatin C","ApoE_pct":14.1,"BBB_est":0.47,"Priority":"HIGH"},
-    {"Formulation":"C12-200-DOPE-PEG","Size_nm":103,"Zeta_mV":-5.2,"CSF_protein":"Albumin (low)","ApoE_pct":9.8,"BBB_est":0.33,"Priority":"MEDIUM"},
-    {"Formulation":"DODAP-DSPC-Chol","Size_nm":118,"Zeta_mV":-2.1,"CSF_protein":"Transferrin","ApoE_pct":7.2,"BBB_est":0.24,"Priority":"LOW"},
-]
-
-# S1-R12c · UVM — GNAQ/GNA11 + Vitreous corona + m6A
-UVM_VARIANTS = [
-    {"Variant":"GNAQ p.Q209L","Freq_pct":46,"Pathway":"PLCβ → PKC → MAPK","Drug_status":"Darovasertib (clinical)","m6A_writer":"METTL3 upregulated"},
-    {"Variant":"GNA11 p.Q209L","Freq_pct":32,"Pathway":"PLCβ → PKC → MAPK","Drug_status":"Darovasertib (clinical)","m6A_writer":"WTAP upregulated"},
-    {"Variant":"BAP1 loss","Freq_pct":47,"Pathway":"Chromatin remodeling → metastasis","Drug_status":"No approved (HDAC trials)","m6A_writer":"FTO overexpressed"},
-    {"Variant":"SF3B1 p.R625H","Freq_pct":19,"Pathway":"Splicing alteration → neoepitopes","Drug_status":"H3B-8800 (clinical)","m6A_writer":"METTL14 altered"},
-    {"Variant":"EIF1AX p.A113_splice","Freq_pct":14,"Pathway":"Translation initiation","Drug_status":"Novel — no drug","m6A_writer":"YTHDF2 suppressed"},
-]
-UVM_VITREOUS_LNP = [
-    {"Formulation":"SM-102-DSPC-Chol-PEG","Vitreal_protein":"Hyaluronan-binding","Size_nm":95,"Zeta_mV":-3.2,"Retention_h":18,"Priority":"HIGH"},
-    {"Formulation":"Lipid-H-DOPE-PEG","Vitreal_protein":"Vitronectin dominant","Size_nm":88,"Zeta_mV":-4.0,"Retention_h":22,"Priority":"HIGH"},
-    {"Formulation":"DOTAP-DSPC-PEG","Vitreal_protein":"Albumin wash-out","Size_nm":112,"Zeta_mV":+2.1,"Retention_h":6,"Priority":"LOW"},
-    {"Formulation":"MC3-DPPC-Chol","Vitreal_protein":"Clusterin-rich","Size_nm":101,"Zeta_mV":-2.8,"Retention_h":14,"Priority":"MEDIUM"},
-]
-
-# S1-R12d · pAML — FLT3-ITD + BM niche corona + Ferroptosis
-PAML_VARIANTS = [
-    {"Variant":"FLT3-ITD","Freq_pct":25,"Pathway":"RTK constitutive activation → JAK/STAT","Drug_status":"Midostaurin (approved)","Ferroptosis":"GPX4 suppressed"},
-    {"Variant":"NPM1 c.860_863dupTCAG","Freq_pct":30,"Pathway":"Nuclear export deregulation","Drug_status":"APR-548 combo (clinical)","Ferroptosis":"SLC7A11 upregulated"},
-    {"Variant":"DNMT3A p.R882H","Freq_pct":18,"Pathway":"Epigenetic dysregulation","Drug_status":"Azacitidine (approved)","Ferroptosis":"ACSL4 altered"},
-    {"Variant":"CEBPA biallelic","Freq_pct":8,"Pathway":"Myeloid differentiation block","Drug_status":"Novel target","Ferroptosis":"NRF2 pathway"},
-    {"Variant":"IDH1/2 mutation","Freq_pct":15,"Pathway":"2-HG oncometabolite → TET2 inhibition","Drug_status":"Enasidenib (approved)","Ferroptosis":"Iron metabolism disrupted"},
-]
-PAML_BM_LNP = [
-    {"Formulation":"ALC-0315-DSPC-Chol-PEG","BM_protein":"ApoE + Clusterin","Size_nm":98,"Zeta_mV":-3.5,"Marrow_uptake_pct":34,"Priority":"HIGH"},
-    {"Formulation":"MC3-DOPE-Chol-PEG","BM_protein":"Fibronectin dominant","Size_nm":105,"Zeta_mV":-4.2,"Marrow_uptake_pct":28,"Priority":"HIGH"},
-    {"Formulation":"DLin-MC3-DPPC","BM_protein":"Vitronectin-rich","Size_nm":91,"Zeta_mV":-2.9,"Marrow_uptake_pct":19,"Priority":"MEDIUM"},
-    {"Formulation":"Cationic-DOTAP-Chol","BM_protein":"Opsonin-heavy","Size_nm":132,"Zeta_mV":+8.1,"Marrow_uptake_pct":8,"Priority":"LOW"},
-]
-
-# ── S1-F · PHYLO-RARE LOGIC ───────────────────────────────────────────────────
+# ---------- Функції для рідкісних раків ----------
 def dipg_variants(sort_by):
     df = pd.DataFrame(DIPG_VARIANTS).sort_values(
         "Freq_pct" if sort_by == "Frequency" else "Drug_status", ascending=False)
@@ -430,31 +442,31 @@ def paml_ferroptosis(variant):
     )
     return summary, Image.open(buf)
 
-# ── HELPERS ───────────────────────────────────────────────────────────────────
+# ---------- Хелпери для UI ----------
 def section_header(code, name, tagline, projects_html):
     return (
-        f"<div style=\'background:{BG};border:1px solid {BORDER};padding:14px 18px;"
-        f"border-radius:8px;margin-bottom:12px;\'>"
-        f"<div style=\'display:flex;align-items:baseline;gap:10px;\'>"
-        f"<span style=\'color:{ACC2};font-size:16px;font-weight:700\'>{code}</span>"
-        f"<span style=\'color:{TXT};font-size:14px;font-weight:600\'>{name}</span>"
-        f"<span style=\'color:{DIM};font-size:12px\'>{tagline}</span></div>"
-        f"<div style=\'margin-top:8px;font-size:12px;color:{DIM}\'>{projects_html}</div>"
+        f"<div style='background:{BG};border:1px solid {BORDER};padding:14px 18px;"
+        f"border-radius:8px;margin-bottom:12px;'>"
+        f"<div style='display:flex;align-items:baseline;gap:10px;'>"
+        f"<span style='color:{ACC2};font-size:16px;font-weight:700'>{code}</span>"
+        f"<span style='color:{TXT};font-size:14px;font-weight:600'>{name}</span>"
+        f"<span style='color:{DIM};font-size:12px'>{tagline}</span></div>"
+        f"<div style='margin-top:8px;font-size:12px;color:{DIM}'>{projects_html}</div>"
         f"</div>"
     )
 
 def proj_badge(code, title, metric=""):
-    m = (f"<span style=\'background:#0f2a3f;color:{ACC2};padding:1px 7px;border-radius:3px;"
-         f"font-size:10px;margin-left:6px\'>{metric}</span>") if metric else ""
+    m = (f"<span style='background:#0f2a3f;color:{ACC2};padding:1px 7px;border-radius:3px;"
+         f"font-size:10px;margin-left:6px'>{metric}</span>") if metric else ""
     return (
-        f"<div style=\'background:{CARD};border-left:3px solid {ACC};"
-        f"padding:8px 12px;border-radius:0 6px 6px 0;margin-bottom:8px;\'>"
-        f"<span style=\'color:{DIM};font-size:11px\'>{code}</span>{m}<br>"
-        f"<span style=\'color:{TXT};font-size:14px;font-weight:600\'>{title}</span>"
+        f"<div style='background:{CARD};border-left:3px solid {ACC};"
+        f"padding:8px 12px;border-radius:0 6px 6px 0;margin-bottom:8px;'>"
+        f"<span style='color:{DIM};font-size:11px'>{code}</span>{m}<br>"
+        f"<span style='color:{TXT};font-size:14px;font-weight:600'>{title}</span>"
         f"</div>"
     )
 
-# ── CSS ───────────────────────────────────────────────────────────────────────
+# ---------- CSS ----------
 css = f"""
 body, .gradio-container {{ background: {BG} !important; color: {TXT} !important; }}
 
@@ -473,7 +485,7 @@ body, .gradio-container {{ background: {BG} !important; color: {TXT} !important;
     background: {BG} !important;
 }}
 
-/* INNER sub-tab bar — individual tools */
+/* INNER sub-tab bar — R1, R2, etc. */
 .inner-tabs .tab-nav button {{
     color: {DIM} !important;
     background: {BG} !important;
@@ -498,12 +510,25 @@ body, .gradio-container {{ background: {BG} !important; color: {TXT} !important;
     padding: 14px !important;
 }}
 
+/* Третій рівень вкладок (a, b) */
+.sub-sub-tabs .tab-nav button {{
+    color: {DIM} !important;
+    background: {CARD} !important;
+    font-size: 11px !important;
+    padding: 3px 8px !important;
+    border-radius: 3px 3px 0 0 !important;
+}}
+.sub-sub-tabs .tab-nav button.selected {{
+    color: {ACC} !important;
+    background: {BG} !important;
+}}
+
 h1, h2, h3 {{ color: {ACC} !important; }}
 .gr-button-primary {{ background: {ACC} !important; border: none !important; }}
 footer {{ display: none !important; }}
 """
 
-# ── LAB MAP HTML ──────────────────────────────────────────────────────────────
+# ---------- MAP HTML ----------
 MAP_HTML = f"""
 <div style="background:{CARD};padding:22px;border-radius:8px;font-family:monospace;
             font-size:13px;line-height:2.0;color:{TXT}">
@@ -514,62 +539,60 @@ MAP_HTML = f"""
 
 <span style="color:{ACC2};font-weight:600">S1-A · PHYLO-GENOMICS</span>
 <span style="color:{DIM}"> — What breaks in DNA</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R1</b> &nbsp;OpenVariant
+&nbsp;&nbsp;&nbsp;├─ <b>S1-A · R1a</b> &nbsp;OpenVariant
 <span style="color:{GRN}"> AUC=0.939 ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R1b</b> Somatic classifier
+&nbsp;&nbsp;&nbsp;├─ <b>S1-A · R1b</b> Somatic classifier
 <span style="color:#f59e0b"> 🔶 In progress</span><br>
-&nbsp;&nbsp;&nbsp;└─ <b>S1-R12a</b> Rare variants (DIPG · UVM)
+&nbsp;&nbsp;&nbsp;└─ <b>S1-A · R2a</b> Rare variants (DIPG · UVM)
 <span style="color:{DIM}"> 🔴 Planned</span><br><br>
 
 <span style="color:{ACC2};font-weight:600">S1-B · PHYLO-RNA</span>
 <span style="color:{DIM}"> — How to silence it via RNA</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R2</b> &nbsp;miRNA silencing (BRCA1/2, TP53)
+&nbsp;&nbsp;&nbsp;├─ <b>S1-B · R1a</b> &nbsp;miRNA silencing (BRCA2)
 <span style="color:{GRN}"> ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R3</b> &nbsp;siRNA synthetic lethal (LUAD · BRCA · COAD)
+&nbsp;&nbsp;&nbsp;├─ <b>S1-B · R2a</b> &nbsp;siRNA synthetic lethal (LUAD · BRCA · COAD)
 <span style="color:{GRN}"> ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R4</b> &nbsp;lncRNA-TREM2 ceRNA network
+&nbsp;&nbsp;&nbsp;├─ <b>S1-B · R3a</b> &nbsp;lncRNA-TREM2 ceRNA network
 <span style="color:{GRN}"> ✅</span><br>
-&nbsp;&nbsp;&nbsp;└─ <b>S1-R4b</b> ASO designer
+&nbsp;&nbsp;&nbsp;└─ <b>S1-B · R3b</b> ASO designer
 <span style="color:{GRN}"> ✅</span><br><br>
 
 <span style="color:{ACC2};font-weight:600">S1-C · PHYLO-DRUG</span>
 <span style="color:{DIM}"> — Which molecule treats it</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R5</b> &nbsp;FGFR3 RNA-directed compounds
+&nbsp;&nbsp;&nbsp;├─ <b>S1-C · R1a</b> &nbsp;FGFR3 RNA-directed compounds
 <span style="color:{GRN}"> ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R5b</b> Synthetic lethal drug mapping
+&nbsp;&nbsp;&nbsp;├─ <b>S1-C · R1b</b> Synthetic lethal drug mapping
 <span style="color:#f59e0b"> 🔶</span><br>
-&nbsp;&nbsp;&nbsp;└─ <b>S1-R13</b> m6A × Ferroptosis × Circadian ⭐
+&nbsp;&nbsp;&nbsp;└─ <b>S1-C · R2a</b> m6A × Ferroptosis × Circadian ⭐
 <span style="color:{DIM}"> 🔴 Frontier</span><br><br>
 
 <span style="color:{ACC2};font-weight:600">S1-D · PHYLO-LNP</span>
 <span style="color:{DIM}"> — How to deliver the drug</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R6</b> &nbsp;LNP corona (serum)
+&nbsp;&nbsp;&nbsp;├─ <b>S1-D · R1a</b> &nbsp;LNP corona (serum)
 <span style="color:{GRN}"> AUC=0.791 ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R7</b> &nbsp;Flow corona — Vroman effect
+&nbsp;&nbsp;&nbsp;├─ <b>S1-D · R2a</b> &nbsp;Flow corona — Vroman effect
 <span style="color:{GRN}"> ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R8</b> &nbsp;LNP brain / BBB / ApoE
+&nbsp;&nbsp;&nbsp;├─ <b>S1-D · R3a</b> &nbsp;LNP brain / BBB / ApoE
 <span style="color:{GRN}"> ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R10</b> AutoCorona NLP
+&nbsp;&nbsp;&nbsp;├─ <b>S1-D · R4a</b> AutoCorona NLP
 <span style="color:{GRN}"> F1=0.71 ✅</span><br>
-&nbsp;&nbsp;&nbsp;└─ <b>S1-R11</b> CSF · Vitreous · Bone Marrow ⭐
+&nbsp;&nbsp;&nbsp;└─ <b>S1-D · R5a</b> CSF · Vitreous · Bone Marrow ⭐
 <span style="color:{DIM}"> 🔴 0 prior studies</span><br><br>
 
 <span style="color:{ACC2};font-weight:600">S1-E · PHYLO-BIOMARKERS</span>
 <span style="color:{DIM}"> — Detect without biopsy</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R9</b> &nbsp;Liquid Biopsy classifier
+&nbsp;&nbsp;&nbsp;├─ <b>S1-E · R1a</b> &nbsp;Liquid Biopsy classifier
 <span style="color:{GRN}"> AUC=0.992* ✅</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R9b</b> Protein panel validator
-<span style="color:#f59e0b"> 🔶</span><br>
-&nbsp;&nbsp;&nbsp;└─ <b>S1-R9c</b> ctDNA gap analysis
-<span style="color:{DIM}"> 🔴</span><br><br>
+&nbsp;&nbsp;&nbsp;└─ <b>S1-E · R1b</b> Protein panel validator
+<span style="color:#f59e0b"> 🔶</span><br><br>
 
 <span style="color:{ACC2};font-weight:600">S1-F · PHYLO-RARE</span>
 <span style="color:{DIM}"> — Where almost nobody has looked yet (&lt;300 cases/yr)</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R12b</b> DIPG toolkit (H3K27M + CSF LNP + Circadian)
+&nbsp;&nbsp;&nbsp;├─ <b>S1-F · R1a</b> DIPG toolkit (H3K27M + CSF LNP + Circadian)
 <span style="color:#f59e0b"> 🔶 In development</span><br>
-&nbsp;&nbsp;&nbsp;├─ <b>S1-R12c</b> UVM toolkit (GNAQ/GNA11 + vitreous + m6A)
+&nbsp;&nbsp;&nbsp;├─ <b>S1-F · R2a</b> UVM toolkit (GNAQ/GNA11 + vitreous + m6A)
 <span style="color:#f59e0b"> 🔶 In development</span><br>
-&nbsp;&nbsp;&nbsp;└─ <b>S1-R12d</b> pAML toolkit (FLT3-ITD + BM niche + ferroptosis)
+&nbsp;&nbsp;&nbsp;└─ <b>S1-F · R3a</b> pAML toolkit (FLT3-ITD + BM niche + ferroptosis)
 <span style="color:#f59e0b"> 🔶 In development</span><br><br>
 
 <span style="color:{DIM};font-size:11px">
@@ -579,9 +602,8 @@ MAP_HTML = f"""
 </div>
 """
 
-# ── UI ────────────────────────────────────────────────────────────────────────
+# ---------- ПОБУДОВА UI ----------
 with gr.Blocks(css=css, title="K R&D Lab · S1 Biomedical") as demo:
-
     gr.Markdown(
         "# 🔬 K R&D Lab · Science Sphere — S1 Biomedical\n"
         "**Oksana Kolisnyk** · [KOSATIKS GROUP](https://kosatiks-group.pp.ua) &nbsp;|&nbsp; "
@@ -590,406 +612,410 @@ with gr.Blocks(css=css, title="K R&D Lab · S1 Biomedical") as demo:
         "*Research only. Not clinical advice.*"
     )
 
-    # ═══════════════════════════════════════════════════════
-    # OUTER TABS — one per PHYLO-* category
-    # ═══════════════════════════════════════════════════════
-    with gr.Tabs(elem_classes="outer-tabs"):
-
-        # ── 🗺️ MAP ─────────────────────────────────────────
+    # Основний рівень вкладок (категорії)
+    with gr.Tabs(elem_classes="outer-tabs") as outer_tabs:
+        # 1. Карта лабораторії
         with gr.TabItem("🗺️ Lab Map"):
             gr.HTML(MAP_HTML)
 
-        # ── 🧬 S1-A · PHYLO-GENOMICS ───────────────────────
+        # 2. S1-A · PHYLO-GENOMICS
         with gr.TabItem("🧬 PHYLO-GENOMICS"):
             gr.HTML(section_header(
                 "S1-A", "PHYLO-GENOMICS", "— What breaks in DNA",
-                "S1-R1 OpenVariant ✅ &nbsp;·&nbsp; S1-R1b Somatic classifier 🔶 &nbsp;·&nbsp; S1-R12a Rare variants 🔴"
+                "R1a OpenVariant ✅ &nbsp;·&nbsp; R1b Somatic classifier 🔶"
             ))
-            with gr.Tabs(elem_classes="inner-tabs"):
-
-                with gr.TabItem("S1-R1 · OpenVariant"):
-                    gr.HTML(proj_badge("S1-A · PHYLO-GENOMICS · S1-R1",
-                                       "OpenVariant — SNV Pathogenicity Classifier", "AUC = 0.939"))
-                    hgvs = gr.Textbox(label="HGVS notation", placeholder="BRCA1:p.R1699Q")
-                    gr.Markdown("**Or enter functional scores manually:**")
-                    with gr.Row():
-                        sift = gr.Slider(0,1,value=0.5,step=0.01,label="SIFT (0=damaging)")
-                        pp   = gr.Slider(0,1,value=0.5,step=0.01,label="PolyPhen-2")
-                        gn   = gr.Slider(0,0.01,value=0.001,step=0.0001,label="gnomAD AF")
-                    b_v = gr.Button("Predict Pathogenicity", variant="primary")
-                    o_v = gr.HTML()
-                    gr.Examples([["BRCA1:p.R1699Q",0.82,0.05,0.0012],
-                                 ["TP53:p.R248W",0.00,1.00,0.0],
-                                 ["BRCA2:p.D2723A",0.01,0.98,0.0]], inputs=[hgvs,sift,pp,gn], cache_examples=False)
-                    b_v.click(predict_variant, [hgvs,sift,pp,gn], o_v)
-
-                with gr.TabItem("S1-R1b · Somatic 🔶"):
-                    gr.HTML(proj_badge("S1-A · PHYLO-GENOMICS · S1-R1b",
-                                       "Somatic Mutation Classifier — BRCA · LUAD panels", "🔶 In progress"))
-                    gr.Markdown("> This module is in active development. Coming in the next release.")
-
-        # ── 🔬 S1-B · PHYLO-RNA ────────────────────────────
+            with gr.Tabs(elem_classes="inner-tabs") as inner_tabs:
+                # R1 · Variant classification
+                with gr.TabItem("R1 · Variant classification"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs") as sub_tabs:
+                        # R1a · OpenVariant
+                        with gr.TabItem("R1a · OpenVariant"):
+                            gr.HTML(proj_badge("S1-A · R1a", "OpenVariant — SNV Pathogenicity Classifier", "AUC=0.939"))
+                            hgvs = gr.Textbox(label="HGVS notation", placeholder="BRCA1:p.R1699Q")
+                            gr.Markdown("**Or enter functional scores manually:**")
+                            with gr.Row():
+                                sift = gr.Slider(0,1,value=0.5,step=0.01,label="SIFT (0=damaging)")
+                                pp   = gr.Slider(0,1,value=0.5,step=0.01,label="PolyPhen-2")
+                                gn   = gr.Slider(0,0.01,value=0.001,step=0.0001,label="gnomAD AF")
+                            b_v = gr.Button("Predict Pathogenicity", variant="primary")
+                            o_v = gr.HTML()
+                            gr.Examples([["BRCA1:p.R1699Q",0.82,0.05,0.0012],
+                                         ["TP53:p.R248W",0.00,1.00,0.0],
+                                         ["BRCA2:p.D2723A",0.01,0.98,0.0]],
+                                        inputs=[hgvs,sift,pp,gn], cache_examples=False)
+                            b_v.click(predict_variant, [hgvs,sift,pp,gn], o_v)
+                        # R1b · Somatic Classifier (в розробці)
+                        with gr.TabItem("R1b · Somatic Classifier 🔶"):
+                            gr.HTML(proj_badge("S1-A · R1b", "Somatic Mutation Classifier — BRCA · LUAD panels", "🔶 In progress"))
+                            gr.Markdown("> This module is in active development. Coming in the next release.")
+
+        # 3. S1-B · PHYLO-RNA
         with gr.TabItem("🔬 PHYLO-RNA"):
             gr.HTML(section_header(
                 "S1-B", "PHYLO-RNA", "— How to silence it via RNA",
-                "S1-R2 miRNA ✅ &nbsp;·&nbsp; S1-R3 siRNA ✅ &nbsp;·&nbsp; S1-R4 lncRNA ✅ &nbsp;·&nbsp; S1-R4b ASO ✅"
+                "R1a miRNA ✅ &nbsp;·&nbsp; R2a siRNA ✅ &nbsp;·&nbsp; R3a lncRNA ✅ &nbsp;·&nbsp; R3b ASO ✅"
             ))
             with gr.Tabs(elem_classes="inner-tabs"):
-
-                with gr.TabItem("S1-R2 · miRNA"):
-                    gr.HTML(proj_badge("S1-B · PHYLO-RNA · S1-R2",
-                                       "miRNA Silencing — BRCA1/2 · TP53 tumor suppressors"))
-                    g1 = gr.Dropdown(["BRCA2","BRCA1","TP53"], value="BRCA2", label="Gene")
-                    b1 = gr.Button("Find miRNAs", variant="primary")
-                    o1 = gr.Dataframe(label="Top 5 downregulated miRNAs")
-                    gr.Examples([["BRCA2"],["BRCA1"],["TP53"]], inputs=[g1])
-                    b1.click(predict_mirna, [g1], o1)
-
-                with gr.TabItem("S1-R3 · siRNA"):
-                    gr.HTML(proj_badge("S1-B · PHYLO-RNA · S1-R3",
-                                       "siRNA Synthetic Lethal — TP53-null · LUAD · BRCA · COAD"))
-                    g2 = gr.Dropdown(["LUAD","BRCA","COAD"], value="LUAD", label="Cancer type")
-                    b2 = gr.Button("Find Targets", variant="primary")
-                    o2 = gr.Dataframe(label="Top 5 synthetic lethal targets")
-                    gr.Examples([["LUAD"],["BRCA"],["COAD"]], inputs=[g2], cache_examples=False)
-                    b2.click(predict_sirna, [g2], o2)
-
-                with gr.TabItem("S1-R4 · lncRNA + ASO"):
-                    gr.HTML(proj_badge("S1-B · PHYLO-RNA · S1-R4 + S1-R4b",
-                                       "lncRNA-TREM2 ceRNA Network + ASO Candidates · Alzheimer neuroinflammation"))
-                    b3 = gr.Button("Load Results", variant="primary")
-                    o3a = gr.Dataframe(label="ceRNA Network (S1-R4)")
-                    o3b = gr.Dataframe(label="ASO Candidates (S1-R4b)")
-                    b3.click(get_lncrna, [], [o3a, o3b])
-
-        # ── 💊 S1-C · PHYLO-DRUG ───────────────────────────
+                # R1 · miRNA silencing
+                with gr.TabItem("R1 · miRNA silencing"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R1a · BRCA2 miRNA"):
+                            gr.HTML(proj_badge("S1-B · R1a", "miRNA Silencing — BRCA2"))
+                            g1 = gr.Dropdown(["BRCA2","BRCA1","TP53"], value="BRCA2", label="Gene")
+                            b1 = gr.Button("Find miRNAs", variant="primary")
+                            o1 = gr.Dataframe(label="Top 5 downregulated miRNAs")
+                            gr.Examples([["BRCA2"],["BRCA1"],["TP53"]], inputs=[g1])
+                            b1.click(predict_mirna, [g1], o1)
+                # R2 · siRNA SL
+                with gr.TabItem("R2 · siRNA SL"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R2a · TP53 siRNA"):
+                            gr.HTML(proj_badge("S1-B · R2a", "siRNA Synthetic Lethal — TP53-null"))
+                            g2 = gr.Dropdown(["LUAD","BRCA","COAD"], value="LUAD", label="Cancer type")
+                            b2 = gr.Button("Find Targets", variant="primary")
+                            o2 = gr.Dataframe(label="Top 5 synthetic lethal targets")
+                            gr.Examples([["LUAD"],["BRCA"],["COAD"]], inputs=[g2], cache_examples=False)
+                            b2.click(predict_sirna, [g2], o2)
+                # R3 · lncRNA + ASO
+                with gr.TabItem("R3 · lncRNA + ASO"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R3a · lncRNA-TREM2"):
+                            gr.HTML(proj_badge("S1-B · R3a", "lncRNA-TREM2 ceRNA Network"))
+                            b3a = gr.Button("Load ceRNA", variant="primary")
+                            o3a = gr.Dataframe(label="ceRNA Network (R3a)")
+                            b3a.click(lambda: pd.DataFrame(CERNA), [], o3a)
+                        with gr.TabItem("R3b · ASO Designer"):
+                            gr.HTML(proj_badge("S1-B · R3b", "ASO Designer"))
+                            b3b = gr.Button("Load ASO Candidates", variant="primary")
+                            o3b = gr.Dataframe(label="ASO Candidates (R3b)")
+                            b3b.click(lambda: pd.DataFrame(ASO), [], o3b)
+
+        # 4. S1-C · PHYLO-DRUG
         with gr.TabItem("💊 PHYLO-DRUG"):
             gr.HTML(section_header(
                 "S1-C", "PHYLO-DRUG", "— Which molecule treats it",
-                "S1-R5 FGFR3 ✅ &nbsp;·&nbsp; S1-R5b SL drug mapping 🔶 &nbsp;·&nbsp; S1-R13 m6A×Ferroptosis×Circadian 🔴⭐"
+                "R1a FGFR3 ✅ &nbsp;·&nbsp; R1b SL drug mapping 🔶 &nbsp;·&nbsp; R2a m6A×Ferroptosis×Circadian 🔴⭐"
             ))
             with gr.Tabs(elem_classes="inner-tabs"):
+                # R1 · RNA-directed drug
+                with gr.TabItem("R1 · RNA-directed drug"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R1a · FGFR3 RNA Drug"):
+                            gr.HTML(proj_badge("S1-C · R1a", "FGFR3 RNA-Directed Drug Discovery", "top score 0.793"))
+                            g4 = gr.Radio(["P1 (hairpin loop)","P10 (G-quadruplex)"],
+                                          value="P1 (hairpin loop)", label="Target pocket")
+                            b4 = gr.Button("Screen Compounds", variant="primary")
+                            o4t = gr.Dataframe(label="Top 5 candidates")
+                            o4p = gr.Image(label="Binding scores")
+                            gr.Examples([["P1 (hairpin loop)"],["P10 (G-quadruplex)"]], inputs=[g4])
+                            b4.click(predict_drug, [g4], [o4t, o4p])
+                        with gr.TabItem("R1b · SL Drug Mapping 🔶"):
+                            gr.HTML(proj_badge("S1-C · R1b", "Synthetic Lethal Drug Mapping", "🔶 In progress"))
+                            gr.Markdown("> In development. Coming soon.")
+                # R2 · Frontier
+                with gr.TabItem("R2 · Frontier"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R2a · m6A×Ferroptosis×Circadian 🔴⭐"):
+                            gr.HTML(proj_badge("S1-C · R2a", "m6A × Ferroptosis × Circadian", "🔴 Frontier"))
+                            gr.Markdown(
+                                "> **Research gap:** This triple intersection has never been studied as an integrated system.\n\n"
+                                "> **Planned datasets:** TCGA-PAAD · GEO m6A atlases · Circadian gene panels\n\n"
+                                "> **Expected timeline:** Q3 2026"
+                            )
 
-                with gr.TabItem("S1-R5 · FGFR3"):
-                    gr.HTML(proj_badge("S1-C · PHYLO-DRUG · S1-R5",
-                                       "FGFR3 RNA-Directed Drug Discovery · ChEMBL screen",
-                                       "top score 0.793"))
-                    g4 = gr.Radio(["P1 (hairpin loop)","P10 (G-quadruplex)"],
-                                  value="P1 (hairpin loop)", label="Target pocket")
-                    b4 = gr.Button("Screen Compounds", variant="primary")
-                    o4t = gr.Dataframe(label="Top 5 candidates")
-                    o4p = gr.Image(label="Binding scores")
-                    gr.Examples([["P1 (hairpin loop)"],["P10 (G-quadruplex)"]], inputs=[g4])
-                    b4.click(predict_drug, [g4], [o4t, o4p])
-
-                with gr.TabItem("S1-R13 · Frontier 🔴⭐"):
-                    gr.HTML(proj_badge("S1-C · PHYLO-DRUG · S1-R13",
-                                       "m6A × Ferroptosis × Circadian — Pan-cancer triad", "🔴 Frontier"))
-                    gr.Markdown(
-                        "> **Research gap:** This triple intersection has never been studied as an integrated system.\n\n"
-                        "> **Planned datasets:** TCGA-PAAD · GEO m6A atlases · Circadian gene panels\n\n"
-                        "> **Expected timeline:** Q3 2026"
-                    )
-
-        # ── 🧪 S1-D · PHYLO-LNP ────────────────────────────
+        # 5. S1-D · PHYLO-LNP
         with gr.TabItem("🧪 PHYLO-LNP"):
             gr.HTML(section_header(
-                "S1-D", "PHYLO-LNP", "— How to deliver the drug to the cell",
-                "S1-R6 Corona ✅ · S1-R7 Flow ✅ · S1-R8 Brain ✅ · S1-R10 NLP ✅ · S1-R11 CSF/BM 🔴⭐"
+                "S1-D", "PHYLO-LNP", "— How to deliver the drug",
+                "R1a Corona ✅ · R2a Flow ✅ · R3a Brain ✅ · R4a NLP ✅ · R5a CSF/BM 🔴⭐"
             ))
             with gr.Tabs(elem_classes="inner-tabs"):
+                # R1 · Serum corona
+                with gr.TabItem("R1 · Serum corona"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R1a · LNP Corona ML"):
+                            gr.HTML(proj_badge("S1-D · R1a", "LNP Protein Corona (Serum)", "AUC=0.791"))
+                            with gr.Row():
+                                sz = gr.Slider(50,300,value=100,step=1,label="Size (nm)")
+                                zt = gr.Slider(-40,10,value=-5,step=1,label="Zeta (mV)")
+                            with gr.Row():
+                                pg = gr.Slider(0,5,value=1.5,step=0.1,label="PEG mol%")
+                                lp = gr.Dropdown(["Ionizable","Cationic","Anionic","Neutral"],value="Ionizable",label="Lipid type")
+                            b6 = gr.Button("Predict", variant="primary"); o6 = gr.Markdown()
+                            gr.Examples([[100,-5,1.5,"Ionizable"],[80,5,0.5,"Cationic"]], inputs=[sz,zt,pg,lp])
+                            b6.click(predict_corona, [sz,zt,pg,lp], o6)
+                # R2 · Flow corona
+                with gr.TabItem("R2 · Flow corona"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R2a · Flow Corona"):
+                            gr.HTML(proj_badge("S1-D · R2a", "Flow Corona — Vroman Effect"))
+                            with gr.Row():
+                                s8  = gr.Slider(50,300,value=100,step=1,label="Size (nm)")
+                                z8  = gr.Slider(-40,10,value=-5,step=1,label="Zeta (mV)")
+                                pg8 = gr.Slider(0,5,value=1.5,step=0.1,label="PEG mol%")
+                            with gr.Row():
+                                ch8 = gr.Dropdown(["Ionizable","Cationic","Anionic","Neutral"],value="Ionizable",label="Charge")
+                                fl8 = gr.Slider(0,40,value=20,step=1,label="Flow cm/s (aorta=40)")
+                            b8 = gr.Button("Model Vroman Effect", variant="primary")
+                            o8t = gr.Markdown(); o8p = gr.Image(label="Kinetics")
+                            gr.Examples([[100,-5,1.5,"Ionizable",40],[150,5,0.5,"Cationic",10]], inputs=[s8,z8,pg8,ch8,fl8])
+                            b8.click(predict_flow, [s8,z8,pg8,ch8,fl8], [o8t,o8p])
+                # R3 · Brain BBB
+                with gr.TabItem("R3 · Brain BBB"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R3a · LNP Brain"):
+                            gr.HTML(proj_badge("S1-D · R3a", "LNP Brain Delivery"))
+                            smi = gr.Textbox(label="Ionizable lipid SMILES",
+                                             value="CC(C)CC(=O)OCC(COC(=O)CC(C)C)OC(=O)CC(C)C")
+                            with gr.Row():
+                                pk  = gr.Slider(4,8,value=6.5,step=0.1,label="pKa")
+                                zt9 = gr.Slider(-20,10,value=-3,step=1,label="Zeta (mV)")
+                            b9 = gr.Button("Predict BBB Crossing", variant="primary")
+                            o9t = gr.Markdown(); o9p = gr.Image(label="Radar profile")
+                            gr.Examples([["CC(C)CC(=O)OCC(COC(=O)CC(C)C)OC(=O)CC(C)C",6.5,-3]], inputs=[smi,pk,zt9])
+                            b9.click(predict_bbb, [smi,pk,zt9], [o9t,o9p])
+                # R4 · NLP
+                with gr.TabItem("R4 · NLP"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R4a · AutoCorona NLP"):
+                            gr.HTML(proj_badge("S1-D · R4a", "AutoCorona NLP", "F1=0.71"))
+                            txt  = gr.Textbox(lines=5,label="Paper abstract",placeholder="Paste abstract here...")
+                            b10  = gr.Button("Extract Data", variant="primary")
+                            o10j = gr.Code(label="Extracted JSON", language="json")
+                            o10f = gr.Textbox(label="Validation flags")
+                            gr.Examples([[
+                                "LNPs composed of MC3, DSPC, Cholesterol (50:10:40 mol%) with 1.5% PEG-DMG. "
+                                "Hydrodynamic diameter was 98 nm, zeta potential -3.2 mV, PDI 0.12. "
+                                "Incubated in human plasma. Corona: albumin, apolipoprotein E, fibrinogen."
+                            ]], inputs=[txt])
+                            b10.click(extract_corona, txt, [o10j, o10f])
+                # R5 · Exotic fluids
+                with gr.TabItem("R5 · Exotic fluids 🔴⭐"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R5a · CSF/Vitreous/BM"):
+                            gr.HTML(proj_badge("S1-D · R5a", "LNP Corona in CSF · Vitreous · Bone Marrow", "🔴 0 prior studies"))
+                            gr.Markdown(
+                                "> **Research gap:** Protein corona has only been characterized in serum/plasma. "
+                                "CSF, vitreous humor, and bone marrow interstitial fluid remain completely unstudied.\n\n"
+                                "> **Target cancers:** DIPG (CSF) · UVM (vitreous) · pAML (bone marrow)\n\n"
+                                "> **Expected timeline:** Q2–Q3 2026"
+                            )
 
-                with gr.TabItem("S1-R6 · Corona"):
-                    gr.HTML(proj_badge("S1-D · PHYLO-LNP · S1-R6",
-                                       "LNP Protein Corona (Serum)", "AUC = 0.791"))
-                    with gr.Row():
-                        sz = gr.Slider(50,300,value=100,step=1,label="Size (nm)")
-                        zt = gr.Slider(-40,10,value=-5,step=1,label="Zeta (mV)")
-                    with gr.Row():
-                        pg = gr.Slider(0,5,value=1.5,step=0.1,label="PEG mol%")
-                        lp = gr.Dropdown(["Ionizable","Cationic","Anionic","Neutral"],value="Ionizable",label="Lipid type")
-                    b6 = gr.Button("Predict", variant="primary"); o6 = gr.Markdown()
-                    gr.Examples([[100,-5,1.5,"Ionizable"],[80,5,0.5,"Cationic"]], inputs=[sz,zt,pg,lp])
-                    b6.click(predict_corona, [sz,zt,pg,lp], o6)
-
-                with gr.TabItem("S1-R7 · Flow"):
-                    gr.HTML(proj_badge("S1-D · PHYLO-LNP · S1-R7",
-                                       "Flow Corona — Vroman Effect · albumin→ApoE kinetics"))
-                    with gr.Row():
-                        s8  = gr.Slider(50,300,value=100,step=1,label="Size (nm)")
-                        z8  = gr.Slider(-40,10,value=-5,step=1,label="Zeta (mV)")
-                        pg8 = gr.Slider(0,5,value=1.5,step=0.1,label="PEG mol%")
-                    with gr.Row():
-                        ch8 = gr.Dropdown(["Ionizable","Cationic","Anionic","Neutral"],value="Ionizable",label="Charge")
-                        fl8 = gr.Slider(0,40,value=20,step=1,label="Flow cm/s (aorta=40)")
-                    b8 = gr.Button("Model Vroman Effect", variant="primary")
-                    o8t = gr.Markdown(); o8p = gr.Image(label="Kinetics")
-                    gr.Examples([[100,-5,1.5,"Ionizable",40],[150,5,0.5,"Cationic",10]], inputs=[s8,z8,pg8,ch8,fl8])
-                    b8.click(predict_flow, [s8,z8,pg8,ch8,fl8], [o8t,o8p])
-
-                with gr.TabItem("S1-R8 · Brain"):
-                    gr.HTML(proj_badge("S1-D · PHYLO-LNP · S1-R8",
-                                       "LNP Brain Delivery — ApoE% + BBB probability"))
-                    smi = gr.Textbox(label="Ionizable lipid SMILES",
-                                     value="CC(C)CC(=O)OCC(COC(=O)CC(C)C)OC(=O)CC(C)C")
-                    with gr.Row():
-                        pk  = gr.Slider(4,8,value=6.5,step=0.1,label="pKa")
-                        zt9 = gr.Slider(-20,10,value=-3,step=1,label="Zeta (mV)")
-                    b9 = gr.Button("Predict BBB Crossing", variant="primary")
-                    o9t = gr.Markdown(); o9p = gr.Image(label="Radar profile")
-                    gr.Examples([["CC(C)CC(=O)OCC(COC(=O)CC(C)C)OC(=O)CC(C)C",6.5,-3]], inputs=[smi,pk,zt9])
-                    b9.click(predict_bbb, [smi,pk,zt9], [o9t,o9p])
-
-                with gr.TabItem("S1-R10 · NLP"):
-                    gr.HTML(proj_badge("S1-D · PHYLO-LNP · S1-R10",
-                                       "AutoCorona NLP — structured data from PMC abstracts", "F1 = 0.71"))
-                    txt  = gr.Textbox(lines=5,label="Paper abstract",placeholder="Paste abstract here...")
-                    b10  = gr.Button("Extract Data", variant="primary")
-                    o10j = gr.Code(label="Extracted JSON", language="json")
-                    o10f = gr.Textbox(label="Validation flags")
-                    gr.Examples([[
-                        "LNPs composed of MC3, DSPC, Cholesterol (50:10:40 mol%) with 1.5% PEG-DMG. "
-                        "Hydrodynamic diameter was 98 nm, zeta potential -3.2 mV, PDI 0.12. "
-                        "Incubated in human plasma. Corona: albumin, apolipoprotein E, fibrinogen."
-                    ]], inputs=[txt])
-                    b10.click(extract_corona, txt, [o10j, o10f])
-
-                with gr.TabItem("S1-R11 · CSF/BM 🔴⭐"):
-                    gr.HTML(proj_badge("S1-D · PHYLO-LNP · S1-R11",
-                                       "LNP Corona in CSF · Vitreous · Bone Marrow", "🔴 0 prior studies"))
-                    gr.Markdown(
-                        "> **Research gap:** Protein corona has only been characterized in serum/plasma. "
-                        "CSF, vitreous humor, and bone marrow interstitial fluid remain completely unstudied.\n\n"
-                        "> **Target cancers:** DIPG (CSF) · UVM (vitreous) · pAML (bone marrow)\n\n"
-                        "> **Expected timeline:** Q2–Q3 2026"
-                    )
-
-        # ── 🩸 S1-E · PHYLO-BIOMARKERS ─────────────────────
+        # 6. S1-E · PHYLO-BIOMARKERS
         with gr.TabItem("🩸 PHYLO-BIOMARKERS"):
             gr.HTML(section_header(
-                "S1-E", "PHYLO-BIOMARKERS", "— Detect cancer without tissue biopsy",
-                "S1-R9 Liquid Biopsy ✅ &nbsp;·&nbsp; S1-R9b Panel validator 🔶 &nbsp;·&nbsp; S1-R9c ctDNA gap 🔴"
+                "S1-E", "PHYLO-BIOMARKERS", "— Detect without biopsy",
+                "R1a Liquid Biopsy ✅ &nbsp;·&nbsp; R1b Protein validator 🔶"
             ))
             with gr.Tabs(elem_classes="inner-tabs"):
-
-                with gr.TabItem("S1-R9 · Liquid Biopsy"):
-                    gr.HTML(proj_badge("S1-E · PHYLO-BIOMARKERS · S1-R9",
-                                       "Liquid Biopsy Classifier — CTHRC1 · FHL2 · LDHA panel",
-                                       "AUC = 0.992*"))
-                    with gr.Row():
-                        p1=gr.Slider(-3,3,value=0,step=0.1,label="CTHRC1")
-                        p2=gr.Slider(-3,3,value=0,step=0.1,label="FHL2")
-                        p3=gr.Slider(-3,3,value=0,step=0.1,label="LDHA")
-                        p4=gr.Slider(-3,3,value=0,step=0.1,label="P4HA1")
-                        p5=gr.Slider(-3,3,value=0,step=0.1,label="SERPINH1")
-                    with gr.Row():
-                        p6=gr.Slider(-3,3,value=0,step=0.1,label="ABCA8")
-                        p7=gr.Slider(-3,3,value=0,step=0.1,label="CA4")
-                        p8=gr.Slider(-3,3,value=0,step=0.1,label="CKB")
-                        p9=gr.Slider(-3,3,value=0,step=0.1,label="NNMT")
-                        p10=gr.Slider(-3,3,value=0,step=0.1,label="CACNA2D2")
-                    b7=gr.Button("Classify", variant="primary")
-                    o7t=gr.HTML(); o7p=gr.Image(label="Feature contributions")
-                    gr.Examples([[2,2,1.5,1.8,1.6,-1,-1.2,-0.8,1.4,-1.1],[0]*10],
-                                inputs=[p1,p2,p3,p4,p5,p6,p7,p8,p9,p10])
-                    b7.click(predict_cancer, [p1,p2,p3,p4,p5,p6,p7,p8,p9,p10], [o7t,o7p])
-
-                with gr.TabItem("S1-R9b · Validator 🔶"):
-                    gr.HTML(proj_badge("S1-E · PHYLO-BIOMARKERS · S1-R9b",
-                                       "Protein Panel Validator — multi-cancer plasma validation", "🔶 In progress"))
-                    gr.Markdown("> Coming next — validates S1-R9 results against GEO plasma proteomics datasets.")
-
-        # ── 📓 JOURNAL ──────────────────────────────────────
-        # ── 🧠 S1-F · PHYLO-RARE ───────────────────────────
+                with gr.TabItem("R1 · Liquid biopsy"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R1a · Liquid Biopsy"):
+                            gr.HTML(proj_badge("S1-E · R1a", "Liquid Biopsy Classifier", "AUC=0.992*"))
+                            with gr.Row():
+                                p1=gr.Slider(-3,3,value=0,step=0.1,label="CTHRC1")
+                                p2=gr.Slider(-3,3,value=0,step=0.1,label="FHL2")
+                                p3=gr.Slider(-3,3,value=0,step=0.1,label="LDHA")
+                                p4=gr.Slider(-3,3,value=0,step=0.1,label="P4HA1")
+                                p5=gr.Slider(-3,3,value=0,step=0.1,label="SERPINH1")
+                            with gr.Row():
+                                p6=gr.Slider(-3,3,value=0,step=0.1,label="ABCA8")
+                                p7=gr.Slider(-3,3,value=0,step=0.1,label="CA4")
+                                p8=gr.Slider(-3,3,value=0,step=0.1,label="CKB")
+                                p9=gr.Slider(-3,3,value=0,step=0.1,label="NNMT")
+                                p10=gr.Slider(-3,3,value=0,step=0.1,label="CACNA2D2")
+                            b7=gr.Button("Classify", variant="primary")
+                            o7t=gr.HTML(); o7p=gr.Image(label="Feature contributions")
+                            gr.Examples([[2,2,1.5,1.8,1.6,-1,-1.2,-0.8,1.4,-1.1],[0]*10],
+                                        inputs=[p1,p2,p3,p4,p5,p6,p7,p8,p9,p10])
+                            b7.click(predict_cancer, [p1,p2,p3,p4,p5,p6,p7,p8,p9,p10], [o7t,o7p])
+                        with gr.TabItem("R1b · Protein Validator 🔶"):
+                            gr.HTML(proj_badge("S1-E · R1b", "Protein Panel Validator", "🔶 In progress"))
+                            gr.Markdown("> Coming next — validates R1a results against GEO plasma proteomics datasets.")
+
+        # 7. S1-F · PHYLO-RARE
         with gr.TabItem("🧠 PHYLO-RARE"):
             gr.HTML(section_header(
                 "S1-F", "PHYLO-RARE", "— Where almost nobody has looked yet",
-                "<b style='color:#ef4444'>⚠️ &lt;300 cases/yr · &lt;5% survival · 0–1 prior studies per gap</b><br>"
-                "S1-R12b DIPG (H3K27M) 🔶 &nbsp;·&nbsp; "
-                "S1-R12c UVM (GNAQ/GNA11) 🔶 &nbsp;·&nbsp; "
-                "S1-R12d pAML (FLT3-ITD) 🔶"
+                "<b style='color:#ef4444'>⚠️ <300 cases/yr · <5% survival · 0–1 prior studies per gap</b><br>"
+                "R1a DIPG 🔶 &nbsp;·&nbsp; R2a UVM 🔶 &nbsp;·&nbsp; R3a pAML 🔶"
             ))
             with gr.Tabs(elem_classes="inner-tabs"):
-
-                with gr.TabItem("S1-R12b · DIPG"):
-                    gr.HTML(proj_badge(
-                        "S1-F · PHYLO-RARE · S1-R12b",
-                        "DIPG Toolkit — H3K27M · CSF LNP · Circadian Biology",
-                        "PBTA · GSE126319"
-                    ))
-                    gr.Markdown(
-                        "> **Why DIPG?** Diffuse Intrinsic Pontine Glioma — median survival 9–11 months. "
-                        "H3K27M oncohistone in **78%** cases. "
-                        "CSF delivery is the only viable route past the brainstem BBB. "
-                        "Circadian disruption (BMAL1 suppression) newly linked — **0 prior LNP studies**."
-                    )
-                    with gr.Tabs(elem_classes="inner-tabs"):
-                        with gr.TabItem("Variants"):
-                            sort_d = gr.Radio(["Frequency", "Drug status"], value="Frequency", label="Sort by")
-                            b_dv   = gr.Button("Load DIPG Variants", variant="primary")
-                            o_dv   = gr.Dataframe(label="H3K27M co-mutations · PBTA/GSE126319")
-                            b_dv.click(dipg_variants, [sort_d], o_dv)
-                        with gr.TabItem("CSF LNP Formulations"):
-                            with gr.Row():
-                                d_peg  = gr.Slider(0.5, 3.0, value=1.5, step=0.1, label="PEG mol%")
-                                d_size = gr.Slider(60, 150, value=90, step=5, label="Target size (nm)")
-                            b_dc  = gr.Button("Rank CSF Formulations", variant="primary")
-                            o_dct = gr.Dataframe(label="CSF LNP ranking")
-                            o_dcp = gr.Image(label="ApoE% in CSF corona")
-                            b_dc.click(dipg_csf, [d_peg, d_size], [o_dct, o_dcp])
-                        with gr.TabItem("Research Gap"):
+                # R1 · DIPG
+                with gr.TabItem("R1 · DIPG"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R1a · DIPG Toolkit"):
+                            gr.HTML(proj_badge("S1-F · R1a", "DIPG Toolkit", "PBTA · GSE126319"))
                             gr.Markdown(
-                                "**Data:** PBTA (n=240) · GSE126319 (n=28) · GTEx circadian genes\n\n"
-                                "| Layer | Known | This study gap |\n"
-                                "|-------|-------|----------------|\n"
-                                "| Genomics | H3K27M freq=78% | H3K27M × BMAL1/CLOCK |\n"
-                                "| Delivery | CED convection | LNP corona **in CSF** |\n"
-                                "| Biology | PRC2 inhibition | Ferroptosis in H3K27M+ DIPG |"
+                                "> **Why DIPG?** Diffuse Intrinsic Pontine Glioma — median survival 9–11 months. "
+                                "H3K27M oncohistone in **78%** cases. "
+                                "CSF delivery is the only viable route past the brainstem BBB. "
+                                "Circadian disruption (BMAL1 suppression) newly linked — **0 prior LNP studies**."
                             )
-
-                with gr.TabItem("S1-R12c · UVM"):
-                    gr.HTML(proj_badge(
-                        "S1-F · PHYLO-RARE · S1-R12c",
-                        "UVM Toolkit — GNAQ/GNA11 · Vitreous Corona · m6A",
-                        "TCGA-UVM n=80"
-                    ))
-                    gr.Markdown(
-                        "> **Why UVM?** Uveal Melanoma — metastatic 5-yr survival **15%**. "
-                        "GNAQ/GNA11 mutations in 78% cases. "
-                        "Vitreous humor protein corona has **never been profiled**. "
-                        "METTL3/WTAP upregulated in GNAQ+ tumors — 0 therapeutic studies."
-                    )
-                    with gr.Tabs(elem_classes="inner-tabs"):
-                        with gr.TabItem("Variants + m6A"):
-                            b_uv = gr.Button("Load UVM Variants", variant="primary")
-                            o_uv = gr.Dataframe(label="GNAQ/GNA11 map · TCGA-UVM")
-                            b_uv.click(uvm_variants, [], o_uv)
-                        with gr.TabItem("Vitreous LNP"):
-                            b_uw  = gr.Button("Rank Vitreous Formulations", variant="primary")
-                            o_uwt = gr.Dataframe(label="Vitreous LNP retention ranking")
-                            o_uwp = gr.Image(label="Retention (hours)")
-                            b_uw.click(uvm_vitreous, [], [o_uwt, o_uwp])
-                        with gr.TabItem("Research Gap"):
+                            with gr.Tabs():
+                                with gr.TabItem("Variants"):
+                                    sort_d = gr.Radio(["Frequency", "Drug status"], value="Frequency", label="Sort by")
+                                    b_dv   = gr.Button("Load DIPG Variants", variant="primary")
+                                    o_dv   = gr.Dataframe(label="H3K27M co-mutations · PBTA/GSE126319")
+                                    b_dv.click(dipg_variants, [sort_d], o_dv)
+                                with gr.TabItem("CSF LNP"):
+                                    with gr.Row():
+                                        d_peg  = gr.Slider(0.5, 3.0, value=1.5, step=0.1, label="PEG mol%")
+                                        d_size = gr.Slider(60, 150, value=90, step=5, label="Target size (nm)")
+                                    b_dc  = gr.Button("Rank CSF Formulations", variant="primary")
+                                    o_dct = gr.Dataframe(label="CSF LNP ranking")
+                                    o_dcp = gr.Image(label="ApoE% in CSF corona")
+                                    b_dc.click(dipg_csf, [d_peg, d_size], [o_dct, o_dcp])
+                                with gr.TabItem("Research Gap"):
+                                    gr.Markdown(
+                                        "**Data:** PBTA (n=240) · GSE126319 (n=28) · GTEx circadian genes\n\n"
+                                        "| Layer | Known | This study gap |\n"
+                                        "|-------|-------|----------------|\n"
+                                        "| Genomics | H3K27M freq=78% | H3K27M × BMAL1/CLOCK |\n"
+                                        "| Delivery | CED convection | LNP corona **in CSF** |\n"
+                                        "| Biology | PRC2 inhibition | Ferroptosis in H3K27M+ DIPG |"
+                                    )
+                # R2 · UVM
+                with gr.TabItem("R2 · UVM"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R2a · UVM Toolkit"):
+                            gr.HTML(proj_badge("S1-F · R2a", "UVM Toolkit", "TCGA-UVM n=80"))
                             gr.Markdown(
-                                "**Data:** TCGA-UVM (n=80) · GEO m6A atlases · Vitreous proteomics\n\n"
-                                "| Layer | Known | This study gap |\n"
-                                "|-------|-------|----------------|\n"
-                                "| Genomics | GNAQ/GNA11 mutations | m6A landscape GNAQ+ vs GNA11+ |\n"
-                                "| Delivery | Intravitreal injection | LNP corona **in vitreous humor** |\n"
-                                "| Biology | PLCβ→PKC→MAPK | GNAQ × METTL3 × YTHDF2 axis |"
+                                "> **Why UVM?** Uveal Melanoma — metastatic 5-yr survival **15%**. "
+                                "GNAQ/GNA11 mutations in 78% cases. "
+                                "Vitreous humor protein corona has **never been profiled**. "
+                                "METTL3/WTAP upregulated in GNAQ+ tumors — 0 therapeutic studies."
                             )
-
-                with gr.TabItem("S1-R12d · pAML"):
-                    gr.HTML(proj_badge(
-                        "S1-F · PHYLO-RARE · S1-R12d",
-                        "pAML Toolkit — FLT3-ITD · Bone Marrow Corona · Ferroptosis",
-                        "TARGET-AML n≈197"
-                    ))
-                    gr.Markdown(
-                        "> **Why pAML?** Pediatric AML — relapse OS **<30%**. "
-                        "FLT3-ITD in 25% cases. "
-                        "Bone marrow niche LNP corona: **never studied**. "
-                        "Ferroptosis–FLT3 intersection: 0 prior studies (FerrDb v2 confirmed)."
-                    )
-                    with gr.Tabs(elem_classes="inner-tabs"):
-                        with gr.TabItem("Ferroptosis Explorer"):
-                            var_sel = gr.Dropdown(
-                                ["FLT3-ITD", "NPM1 c.860_863dupTCAG", "DNMT3A p.R882H",
-                                 "CEBPA biallelic", "IDH1/2 mutation"],
-                                value="FLT3-ITD", label="Select variant"
-                            )
-                            b_pf  = gr.Button("Analyze Ferroptosis Profile", variant="primary")
-                            o_pft = gr.HTML()
-                            o_pfp = gr.Image(label="Target radar")
-                            b_pf.click(paml_ferroptosis, var_sel, [o_pft, o_pfp])
-                        with gr.TabItem("BM Niche LNP"):
-                            gr.Dataframe(
-                                value=pd.DataFrame(PAML_BM_LNP),
-                                label="Bone marrow niche LNP candidates · TARGET-AML context"
-                            )
-                        with gr.TabItem("Research Gap"):
+                            with gr.Tabs():
+                                with gr.TabItem("Variants + m6A"):
+                                    b_uv = gr.Button("Load UVM Variants", variant="primary")
+                                    o_uv = gr.Dataframe(label="GNAQ/GNA11 map · TCGA-UVM")
+                                    b_uv.click(uvm_variants, [], o_uv)
+                                with gr.TabItem("Vitreous LNP"):
+                                    b_uw  = gr.Button("Rank Vitreous Formulations", variant="primary")
+                                    o_uwt = gr.Dataframe(label="Vitreous LNP retention ranking")
+                                    o_uwp = gr.Image(label="Retention (hours)")
+                                    b_uw.click(uvm_vitreous, [], [o_uwt, o_uwp])
+                                with gr.TabItem("Research Gap"):
+                                    gr.Markdown(
+                                        "**Data:** TCGA-UVM (n=80) · GEO m6A atlases · Vitreous proteomics\n\n"
+                                        "| Layer | Known | This study gap |\n"
+                                        "|-------|-------|----------------|\n"
+                                        "| Genomics | GNAQ/GNA11 mutations | m6A landscape GNAQ+ vs GNA11+ |\n"
+                                        "| Delivery | Intravitreal injection | LNP corona **in vitreous humor** |\n"
+                                        "| Biology | PLCβ→PKC→MAPK | GNAQ × METTL3 × YTHDF2 axis |"
+                                    )
+                # R3 · pAML
+                with gr.TabItem("R3 · pAML"):
+                    with gr.Tabs(elem_classes="sub-sub-tabs"):
+                        with gr.TabItem("R3a · pAML Toolkit"):
+                            gr.HTML(proj_badge("S1-F · R3a", "pAML Toolkit", "TARGET-AML n≈197"))
                             gr.Markdown(
-                                "**Data:** TARGET-AML (n=197) · BeatAML · FerrDb v2\n\n"
-                                "| Layer | Known | This study gap |\n"
-                                "|-------|-------|----------------|\n"
-                                "| Genomics | FLT3-ITD → Midostaurin | FLT3-ITD × GPX4/SLC7A11 |\n"
-                                "| Delivery | Liposomal daunorubicin | LNP corona **in bone marrow** |\n"
-                                "| Biology | Midostaurin inhibits FLT3 | Ferroptosis SL + FLT3i |"
+                                "> **Why pAML?** Pediatric AML — relapse OS **<30%**. "
+                                "FLT3-ITD in 25% cases. "
+                                "Bone marrow niche LNP corona: **never studied**. "
+                                "Ferroptosis–FLT3 intersection: 0 prior studies (FerrDb v2 confirmed)."
                             )
-
+                            with gr.Tabs():
+                                with gr.TabItem("Ferroptosis Explorer"):
+                                    var_sel = gr.Dropdown(
+                                        ["FLT3-ITD", "NPM1 c.860_863dupTCAG", "DNMT3A p.R882H",
+                                         "CEBPA biallelic", "IDH1/2 mutation"],
+                                        value="FLT3-ITD", label="Select variant"
+                                    )
+                                    b_pf  = gr.Button("Analyze Ferroptosis Profile", variant="primary")
+                                    o_pft = gr.HTML()
+                                    o_pfp = gr.Image(label="Target radar")
+                                    b_pf.click(paml_ferroptosis, var_sel, [o_pft, o_pfp])
+                                with gr.TabItem("BM Niche LNP"):
+                                    gr.Dataframe(
+                                        value=pd.DataFrame(PAML_BM_LNP),
+                                        label="Bone marrow niche LNP candidates · TARGET-AML context"
+                                    )
+                                with gr.TabItem("Research Gap"):
+                                    gr.Markdown(
+                                        "**Data:** TARGET-AML (n=197) · BeatAML · FerrDb v2\n\n"
+                                        "| Layer | Known | This study gap |\n"
+                                        "|-------|-------|----------------|\n"
+                                        "| Genomics | FLT3-ITD → Midostaurin | FLT3-ITD × GPX4/SLC7A11 |\n"
+                                        "| Delivery | Liposomal daunorubicin | LNP corona **in bone marrow** |\n"
+                                        "| Biology | Midostaurin inhibits FLT3 | Ferroptosis SL + FLT3i |"
+                                    )
+
+        # 8. Journal
         with gr.TabItem("📓 Journal"):
             gr.Markdown("### Lab Journal\nEvery tool call auto-logged with project code.")
             with gr.Row():
                 note_text = gr.Textbox(label="📝 Observation", placeholder="What did you discover?", lines=3)
-                note_tab  = gr.Textbox(label="Project code (e.g. S1-R1)", value="General")
+                note_tab  = gr.Textbox(label="Project code (e.g. S1-A-R1a)", value="General")
             note_last  = gr.Textbox(visible=False)
             save_btn   = gr.Button("💾 Save", variant="primary")
             save_msg   = gr.Markdown()
             journal_df = gr.Dataframe(label="📋 Full History", value=load_journal(), interactive=False)
             refresh_btn = gr.Button("🔄 Refresh")
             refresh_btn.click(load_journal, [], journal_df)
-            save_btn.click(save_note, [note_text,note_tab,note_last], [save_msg,journal_df])
+            save_btn.click(save_note, [note_text, note_tab, note_last], [save_msg, journal_df])
 
-        # ── 📚 LEARNING ─────────────────────────────────────
+        # 9. Learning
         with gr.TabItem("📚 Learning"):
             gr.Markdown("""
 ## 🧪 Guided Investigations — S1 Biomedical
 > 🟢 Beginner → 🟡 Intermediate → 🔴 Advanced
 
 ---
-### 🟢 Case 1 · S1-A · S1-R1
+### 🟢 Case 1 · S1-A · R1a
 **Why does the same position give two different outcomes?**
-1. PHYLO-GENOMICS → OpenVariant → `BRCA1:p.R1699Q` → Benign
+1. PHYLO-GENOMICS → R1a · OpenVariant → `BRCA1:p.R1699Q` → Benign
 2. Enter `BRCA1:p.R1699W` → Pathogenic
 3. Same position, different amino acid — Q (polar) vs W (bulky-aromatic)
 
 ---
-### 🟢 Case 2 · S1-D · S1-R6 + S1-R8
+### 🟢 Case 2 · S1-D · R1a + R3a
 **How does PEG% control which protein coats the nanoparticle?**
-1. PHYLO-LNP → Corona → Ionizable, Zeta=−5, PEG=**0.5%** → note protein
+1. PHYLO-LNP → R1a · Corona → Ionizable, Zeta=−5, PEG=**0.5%** → note protein
 2. Change PEG=**2.5%** → compare
-3. Brain tab → pKa=6.5 → check ApoE%
+3. R3a · Brain → pKa=6.5 → check ApoE%
 
 ---
-### 🟡 Case 3 · S1-D · S1-R7
+### 🟡 Case 3 · S1-D · R2a
 **Does blood flow reshape the corona over time?**
-1. PHYLO-LNP → Flow → Flow=0 → observe ApoE curve
+1. PHYLO-LNP → R2a · Flow → Flow=0 → observe ApoE curve
 2. Flow=40 (arterial) → compare
 3. At what minute does ApoE dominate?
 
 ---
-### 🟡 Case 4 · S1-B · S1-R3
+### 🟡 Case 4 · S1-B · R2a
 **Which cancer has the most novel (undrugged) siRNA targets?**
-1. PHYLO-RNA → siRNA → LUAD → count "Novel"
+1. PHYLO-RNA → R2a · siRNA → LUAD → count "Novel"
 2. Repeat BRCA, COAD
 
 ---
-### 🔴 Case 5 · S1-E · S1-R9
+### 🔴 Case 5 · S1-E · R1a
 **Minimum protein signal that flips to CANCER?**
-1. PHYLO-BIOMARKERS → Liquid Biopsy → all=0 → HEALTHY
+1. PHYLO-BIOMARKERS → R1a · Liquid Biopsy → all=0 → HEALTHY
 2. Set CTHRC1=2.5, FHL2=2.0, LDHA=1.8 → observe
 3. Reset. Increase only CTHRC1 step by step.
 
 ---
-### 🔴 Case 6 · S1-B + S1-C — Cross-tool convergence
-1. PHYLO-RNA → miRNA → TP53 → find top targets (BCL2, CDK6)
-2. PHYLO-DRUG → FGFR3 → check CDK6 pathway overlap
-3. PHYLO-RNA → siRNA → BRCA → does CDK6 appear?
+### 🔴 Case 6 · Cross-tool convergence
+1. PHYLO-RNA → R1a · miRNA → TP53 → find top targets (BCL2, CDK6)
+2. PHYLO-DRUG → R1a · FGFR3 → check CDK6 pathway overlap
+3. PHYLO-RNA → R2a · siRNA → BRCA → does CDK6 appear?
 
 ---
 ### 📖 Tool Index
 | Code | Module | Tool | Metric |
 |------|--------|------|--------|
-| S1-R1 | PHYLO-GENOMICS | OpenVariant | AUC=0.939 |
-| S1-R2 | PHYLO-RNA | miRNA silencing | top: hsa-miR-148a |
-| S1-R3 | PHYLO-RNA | siRNA SL | SPC24 top LUAD |
-| S1-R4 | PHYLO-RNA | lncRNA-TREM2 | CYTOR→AKT1 |
-| S1-R5 | PHYLO-DRUG | FGFR3 drug | score=0.793 |
-| S1-R6 | PHYLO-LNP | Corona | AUC=0.791 |
-| S1-R7 | PHYLO-LNP | Flow Vroman | 3–4× faster |
-| S1-R8 | PHYLO-LNP | LNP Brain | pKa 6.2–6.8 |
-| S1-R9 | PHYLO-BIOMARKERS | Liquid Biopsy | AUC=0.992* |
-| S1-R10 | PHYLO-LNP | AutoCorona NLP | F1=0.71 |
+| S1-A·R1a | PHYLO-GENOMICS | OpenVariant | AUC=0.939 |
+| S1-B·R1a | PHYLO-RNA | miRNA silencing | top: hsa-miR-148a |
+| S1-B·R2a | PHYLO-RNA | siRNA SL | SPC24 top LUAD |
+| S1-B·R3a | PHYLO-RNA | lncRNA-TREM2 | CYTOR→AKT1 |
+| S1-C·R1a | PHYLO-DRUG | FGFR3 drug | score=0.793 |
+| S1-D·R1a | PHYLO-LNP | Corona | AUC=0.791 |
+| S1-D·R2a | PHYLO-LNP | Flow Vroman | 3–4× faster |
+| S1-D·R3a | PHYLO-LNP | LNP Brain | pKa 6.2–6.8 |
+| S1-E·R1a | PHYLO-BIOMARKERS | Liquid Biopsy | AUC=0.992* |
+| S1-D·R4a | PHYLO-LNP | AutoCorona NLP | F1=0.71 |
 """)
 
     gr.Markdown(