""" Operon Diffusion — Morphogen Gradient Visualizer (Gradio Demo) ============================================================== Three-tab demo: 1. Linear Chain — gradient formation on a line graph 2. Topologies — preset graph shapes (star, ring, grid, tree) 3. Competing Sources — two morphogens diffusing simultaneously Run locally: pip install gradio python space-diffusion/app.py """ import sys from pathlib import Path import gradio as gr # Allow importing operon_ai from the repo root when running locally _repo_root = Path(__file__).resolve().parent.parent if str(_repo_root) not in sys.path: sys.path.insert(0, str(_repo_root)) from operon_ai import ( MorphogenType, MorphogenSource, DiffusionParams, DiffusionField, ) # ── Topology builders ──────────────────────────────────────────────────── NODE_COLORS = [ "#6366f1", "#3b82f6", "#22c55e", "#eab308", "#f97316", "#ef4444", "#ec4899", "#8b5cf6", ] def _build_linear(n: int) -> dict[str, list[str]]: """A → B → C → ... linear chain.""" nodes = [chr(65 + i) for i in range(n)] adj: dict[str, list[str]] = {nd: [] for nd in nodes} for i in range(n - 1): adj[nodes[i]].append(nodes[i + 1]) adj[nodes[i + 1]].append(nodes[i]) return adj def _build_star(n: int) -> dict[str, list[str]]: """Hub 'A' connected to B, C, D, ...""" nodes = [chr(65 + i) for i in range(n)] adj: dict[str, list[str]] = {nd: [] for nd in nodes} hub = nodes[0] for spoke in nodes[1:]: adj[hub].append(spoke) adj[spoke].append(hub) return adj def _build_ring(n: int) -> dict[str, list[str]]: """A → B → C → ... → A cycle.""" nodes = [chr(65 + i) for i in range(n)] adj: dict[str, list[str]] = {nd: [] for nd in nodes} for i in range(n): nxt = (i + 1) % n adj[nodes[i]].append(nodes[nxt]) adj[nodes[nxt]].append(nodes[i]) # Deduplicate for nd in adj: adj[nd] = list(dict.fromkeys(adj[nd])) return adj def _build_grid() -> dict[str, list[str]]: """2x3 grid: A-B-C / D-E-F.""" adj: dict[str, list[str]] = { "A": ["B", "D"], "B": ["A", "C", "E"], "C": ["B", "F"], "D": ["A", "E"], "E": ["B", "D", "F"], "F": ["C", "E"], } return adj def _build_binary_tree() -> dict[str, list[str]]: """A root, B/C children, D/E under B, F/G under C.""" adj: dict[str, list[str]] = { "A": ["B", "C"], "B": ["A", "D", "E"], "C": ["A", "F", "G"], "D": ["B"], "E": ["B"], "F": ["C"], "G": ["C"], } return adj TOPOLOGIES = { "Linear": lambda: _build_linear(5), "Star": lambda: _build_star(6), "Ring": lambda: _build_ring(5), "Grid (2x3)": _build_grid, "Binary Tree": _build_binary_tree, } # ── Visualization helpers ──────────────────────────────────────────────── def _concentration_bars(snapshot: dict[str, dict[str, float]], morphogen_filter: str | None = None) -> str: """Render concentration bars per node.""" if not snapshot: return "

No data yet.

" rows = [] for i, (node_id, concs) in enumerate(sorted(snapshot.items())): color = NODE_COLORS[i % len(NODE_COLORS)] node_rows = [] if morphogen_filter: items = [(morphogen_filter, concs.get(morphogen_filter, 0.0))] else: items = sorted(concs.items()) if concs else [("(none)", 0.0)] for mtype, val in items: pct = max(0, min(100, val * 100)) node_rows.append( f'
' f'{mtype}' f'
' f'
' f'{val:.4f}' f'
' ) rows.append( f'
' f'Node {node_id}' + "".join(node_rows) + "
" ) return '
' + "".join(rows) + "
" def _snapshot_table(snapshot: dict[str, dict[str, float]]) -> str: """Render snapshot as a markdown table.""" if not snapshot: return "No data." # Collect all morphogen types all_types = sorted({mt for concs in snapshot.values() for mt in concs}) if not all_types: all_types = ["(empty)"] header = "| Node | " + " | ".join(all_types) + " |" sep = "| :--- | " + " | ".join("---:" for _ in all_types) + " |" rows = [header, sep] for node_id in sorted(snapshot.keys()): concs = snapshot[node_id] vals = " | ".join(f"{concs.get(mt, 0.0):.4f}" for mt in all_types) rows.append(f"| {node_id} | {vals} |") return "\n".join(rows) def _gradient_level_table(field: DiffusionField, nodes: list[str]) -> str: """Show MorphogenGradient level per node via get_local_gradient().""" rows = ["| Node | Morphogen | Concentration | Level |", "| :--- | :--- | ---: | :--- |"] for nd in sorted(nodes): gradient = field.get_local_gradient(nd) for mt in MorphogenType: val = gradient.get(mt) if val > 0: level = gradient.get_level(mt) rows.append(f"| {nd} | {mt.value} | {val:.4f} | {level} |") if len(rows) == 2: return "No non-zero concentrations to display." return "\n".join(rows) # ── Tab 1: Linear Chain ───────────────────────────────────────────────── def run_linear_chain( num_nodes: int, source_pos: str, emission_rate: float, diffusion_rate: float, decay_rate: float, num_steps: int, ) -> tuple[str, str]: """Run diffusion on a linear chain. Returns (bars_html, step_snapshots_md). """ n = int(num_nodes) nodes = [chr(65 + i) for i in range(n)] adj = _build_linear(n) params = DiffusionParams( diffusion_rate=float(diffusion_rate), decay_rate=float(decay_rate), ) field = DiffusionField.from_adjacency(adj, params=params) source_node = source_pos if source_pos in nodes else nodes[0] field.add_source(MorphogenSource( node_id=source_node, morphogen_type=MorphogenType.COMPLEXITY, emission_rate=float(emission_rate), )) steps = int(num_steps) step_rows = ["| Step | " + " | ".join(nodes) + " |"] step_rows.append("| ---: | " + " | ".join("---:" for _ in nodes) + " |") for s in range(1, steps + 1): field.step() snap = field.snapshot() vals = " | ".join( f"{snap.get(nd, {}).get('complexity', 0.0):.4f}" for nd in nodes ) step_rows.append(f"| {s} | {vals} |") bars_html = _concentration_bars(field.snapshot(), morphogen_filter="complexity") steps_md = "\n".join(step_rows) return bars_html, steps_md # ── Tab 2: Topologies ─────────────────────────────────────────────────── def run_topology( topo_name: str, source_node: str, emission_rate: float, diffusion_rate: float, decay_rate: float, num_steps: int, ) -> tuple[str, str, str]: """Run diffusion on a preset topology. Returns (bars_html, snapshot_md, gradient_md). """ builder = TOPOLOGIES.get(topo_name, TOPOLOGIES["Linear"]) adj = builder() params = DiffusionParams( diffusion_rate=float(diffusion_rate), decay_rate=float(decay_rate), ) field = DiffusionField.from_adjacency(adj, params=params) nodes = sorted(adj.keys()) src = source_node if source_node in nodes else nodes[0] field.add_source(MorphogenSource( node_id=src, morphogen_type=MorphogenType.COMPLEXITY, emission_rate=float(emission_rate), )) field.run(int(num_steps)) bars_html = _concentration_bars(field.snapshot(), morphogen_filter="complexity") snap_md = "### Concentration Snapshot\n\n" + _snapshot_table(field.snapshot()) grad_md = "### Local Gradient Levels\n\n" + _gradient_level_table(field, nodes) return bars_html, snap_md, grad_md def _get_topology_nodes(topo_name: str) -> dict: """Return dropdown update with nodes for the selected topology.""" builder = TOPOLOGIES.get(topo_name, TOPOLOGIES["Linear"]) adj = builder() nodes = sorted(adj.keys()) return gr.update(choices=nodes, value=nodes[0]) # ── Tab 3: Competing Sources ──────────────────────────────────────────── def run_competing( num_nodes: int, src1_pos: str, src1_type: str, src1_rate: float, src2_pos: str, src2_type: str, src2_rate: float, diffusion_rate: float, decay_rate: float, num_steps: int, ) -> tuple[str, str]: """Two morphogens competing on a linear chain. Returns (bars_html, table_md). """ n = int(num_nodes) nodes = [chr(65 + i) for i in range(n)] adj = _build_linear(n) params = DiffusionParams( diffusion_rate=float(diffusion_rate), decay_rate=float(decay_rate), ) field = DiffusionField.from_adjacency(adj, params=params) mt_map = {mt.value: mt for mt in MorphogenType} mt1 = mt_map.get(src1_type, MorphogenType.COMPLEXITY) mt2 = mt_map.get(src2_type, MorphogenType.CONFIDENCE) s1 = src1_pos if src1_pos in nodes else nodes[0] s2 = src2_pos if src2_pos in nodes else nodes[-1] field.add_source(MorphogenSource(node_id=s1, morphogen_type=mt1, emission_rate=float(src1_rate))) field.add_source(MorphogenSource(node_id=s2, morphogen_type=mt2, emission_rate=float(src2_rate))) field.run(int(num_steps)) bars_html = _concentration_bars(field.snapshot()) # Build per-morphogen table snap = field.snapshot() types_present = sorted({mt for concs in snap.values() for mt in concs}) header = "| Node | " + " | ".join(types_present) + " |" sep = "| :--- | " + " | ".join("---:" for _ in types_present) + " |" rows = [header, sep] for nd in nodes: concs = snap.get(nd, {}) vals = " | ".join(f"{concs.get(t, 0.0):.4f}" for t in types_present) rows.append(f"| {nd} | {vals} |") table_md = "### Concentration per Morphogen\n\n" + "\n".join(rows) return bars_html, table_md # ── Gradio UI ──────────────────────────────────────────────────────────── def build_app() -> gr.Blocks: with gr.Blocks(title="Diffusion Visualizer") as app: gr.Markdown( "# 🌊 Diffusion — Morphogen Gradient Visualizer\n" "Simulate **morphogen diffusion** on graph topologies and watch " "concentration gradients form step by step." ) with gr.Tabs(): # ── Tab 1: Linear Chain ────────────────────────────────── with gr.TabItem("Linear Chain"): gr.Markdown( "Emit morphogen from one node in a **linear chain** and " "watch the gradient decay with distance." ) with gr.Row(): lc_nodes = gr.Slider(3, 8, value=5, step=1, label="Number of Nodes") lc_source = gr.Dropdown( choices=[chr(65 + i) for i in range(8)], value="A", label="Source Node", ) with gr.Row(): lc_emission = gr.Slider(0.05, 1.0, value=0.5, step=0.05, label="Emission Rate") lc_diffusion = gr.Slider(0.01, 0.5, value=0.1, step=0.01, label="Diffusion Rate") lc_decay = gr.Slider(0.0, 0.3, value=0.05, step=0.01, label="Decay Rate") with gr.Row(): lc_steps = gr.Slider(1, 50, value=10, step=1, label="Number of Steps") lc_btn = gr.Button("Run Diffusion", variant="primary") lc_bars = gr.HTML(label="Final Concentrations") lc_timeline = gr.Markdown(label="Step-by-Step Snapshot") lc_btn.click( fn=run_linear_chain, inputs=[lc_nodes, lc_source, lc_emission, lc_diffusion, lc_decay, lc_steps], outputs=[lc_bars, lc_timeline], ) # ── Tab 2: Topologies ──────────────────────────────────── with gr.TabItem("Topologies"): gr.Markdown( "Choose a **graph topology** and see how shape affects " "gradient formation. The local gradient level bridges to " "the `MorphogenGradient` API." ) with gr.Row(): tp_topo = gr.Dropdown( choices=list(TOPOLOGIES.keys()), value="Star", label="Topology", ) tp_source = gr.Dropdown( choices=sorted(_build_star(6).keys()), value="A", label="Source Node", ) with gr.Row(): tp_emission = gr.Slider(0.05, 1.0, value=0.5, step=0.05, label="Emission Rate") tp_diffusion = gr.Slider(0.01, 0.5, value=0.1, step=0.01, label="Diffusion Rate") tp_decay = gr.Slider(0.0, 0.3, value=0.05, step=0.01, label="Decay Rate") with gr.Row(): tp_steps = gr.Slider(1, 50, value=15, step=1, label="Number of Steps") tp_btn = gr.Button("Run Diffusion", variant="primary") tp_bars = gr.HTML(label="Concentration Bars") with gr.Row(): with gr.Column(): tp_snap = gr.Markdown(label="Snapshot Table") with gr.Column(): tp_grad = gr.Markdown(label="Gradient Levels") tp_topo.change( fn=_get_topology_nodes, inputs=[tp_topo], outputs=[tp_source], ) tp_btn.click( fn=run_topology, inputs=[tp_topo, tp_source, tp_emission, tp_diffusion, tp_decay, tp_steps], outputs=[tp_bars, tp_snap, tp_grad], ) # ── Tab 3: Competing Sources ───────────────────────────── with gr.TabItem("Competing Sources"): gr.Markdown( "Place **two different morphogens** at different nodes " "and observe overlapping gradients. Each morphogen type " "diffuses independently." ) with gr.Row(): cs_nodes = gr.Slider(3, 8, value=5, step=1, label="Chain Length") morphogen_choices = [mt.value for mt in MorphogenType] with gr.Row(): with gr.Column(): gr.Markdown("#### Source 1") cs_src1 = gr.Dropdown( choices=[chr(65 + i) for i in range(8)], value="A", label="Node", ) cs_type1 = gr.Dropdown( choices=morphogen_choices, value="complexity", label="Morphogen Type", ) cs_rate1 = gr.Slider(0.05, 1.0, value=0.5, step=0.05, label="Emission Rate") with gr.Column(): gr.Markdown("#### Source 2") cs_src2 = gr.Dropdown( choices=[chr(65 + i) for i in range(8)], value="E", label="Node", ) cs_type2 = gr.Dropdown( choices=morphogen_choices, value="confidence", label="Morphogen Type", ) cs_rate2 = gr.Slider(0.05, 1.0, value=0.3, step=0.05, label="Emission Rate") with gr.Row(): cs_diffusion = gr.Slider(0.01, 0.5, value=0.1, step=0.01, label="Diffusion Rate") cs_decay = gr.Slider(0.0, 0.3, value=0.05, step=0.01, label="Decay Rate") cs_steps = gr.Slider(1, 50, value=15, step=1, label="Steps") cs_btn = gr.Button("Run Competing Diffusion", variant="primary") cs_bars = gr.HTML(label="Concentration Bars") cs_table = gr.Markdown(label="Per-Morphogen Table") cs_btn.click( fn=run_competing, inputs=[ cs_nodes, cs_src1, cs_type1, cs_rate1, cs_src2, cs_type2, cs_rate2, cs_diffusion, cs_decay, cs_steps, ], outputs=[cs_bars, cs_table], ) return app if __name__ == "__main__": app = build_app() app.launch(theme=gr.themes.Soft())