Feat/visualization module

requirements.txt

@@ -0,0 +1,5 @@
+dash
+plotly
+umap-learn
+scikit-learn
+matplotlib

src/visualization/flexvis.py

@@ -0,0 +1,156 @@
+import matplotlib
+matplotlib.use("Agg")  # Non-interactive backend
+import matplotlib.pyplot as plt
+import numpy as np
+import io, base64
+
+from dash import Dash, html, dcc, Input, Output
+from viz_flexible_interactive import CombinedPolicy, REG
+
+# -------------------------
+# Helper functions
+# -------------------------
+def fig_to_base64(fig):
+    buf = io.BytesIO()
+    fig.savefig(buf, format="png", bbox_inches="tight")
+    plt.close(fig)
+    buf.seek(0)
+    return "data:image/png;base64," + base64.b64encode(buf.read()).decode("utf-8")
+
+def plot_heat_mpl_dash(X, row_labels=None, col_labels=None):
+    fig, ax = plt.subplots(figsize=(8,6))
+    im = ax.imshow(X, aspect="auto", interpolation="nearest", cmap="viridis")
+    if row_labels:
+        ax.set_yticks(range(len(row_labels)))
+        ax.set_yticklabels(row_labels, fontsize=8)
+    if col_labels:
+        ax.set_xticks(range(len(col_labels)))
+        ax.set_xticklabels(col_labels, fontsize=8, rotation=90)
+    fig.colorbar(im, ax=ax)
+    fig.tight_layout()
+    return fig_to_base64(fig)
+
+def plot_umap_mpl_dash(emb, labels=None):
+    fig, ax = plt.subplots(figsize=(8,6))
+    if labels is None:
+        labels = np.arange(len(emb))
+    scatter = ax.scatter(emb[:,0], emb[:,1], c=labels, cmap="tab20")
+    fig.colorbar(scatter, ax=ax, label="Label")
+    ax.set_xlabel("UMAP1")
+    ax.set_ylabel("UMAP2")
+    ax.set_title("UMAP")
+    fig.tight_layout()
+    return fig_to_base64(fig)
+
+def mpl_to_dash_figure(base64_img):
+    return {
+        "data": [],
+        "layout": {
+            "images": [{
+                "source": base64_img,
+                "xref": "paper",
+                "yref": "paper",
+                "x": 0, "y": 1,
+                "sizex": 1, "sizey": 1,
+                "xanchor": "left", "yanchor": "top",
+                "layer": "below"
+            }],
+            "xaxis": {"visible": False},
+            "yaxis": {"visible": False},
+            "margin": {"l":0, "r":0, "t":0, "b":0},
+        }
+    }
+
+# -------------------------
+# Dummy data
+# -------------------------
+rng = np.random.default_rng(42)
+X = rng.normal(size=(40, 25)) * 2 + rng.normal(size=(40, 1)) * 3
+rows = [f"Gene_{i}" for i in range(X.shape[0])]
+cols = [f"Cell_{j}" for j in range(X.shape[1])]
+point_labels = [f"C{i}" for i in range(X.shape[0])]
+
+# -------------------------
+# Dash app
+# -------------------------
+app = Dash(__name__)
+app.layout = html.Div([
+    html.Div([
+        html.Label("Heatmap backend"),
+        dcc.Dropdown(["matplotlib", "plotly"], "plotly", id="heat-backend"),
+        html.Label("UMAP backend"),
+        dcc.Dropdown(["matplotlib", "plotly_express"], "plotly_express", id="umap-backend"),
+        html.Label("Normalizer"),
+        dcc.Dropdown(list(REG.normalizers.keys()), "zscore_rows", id="normalizer"),
+        html.Label("Clustering"),
+        dcc.Dropdown(list(REG.clusterers.keys()), "simple", id="clusterer"),
+        html.Label("Colormap scale"),
+        dcc.Dropdown(list(REG.cmap_scales.keys()), "center_zero", id="cmap_scale"),
+        html.Label("Pre-UMAP PCA components"),
+        dcc.Slider(min=0, max=25, step=1, value=10,
+                   marks={i:str(i) for i in range(0,26)}, id="pca-components")
+    ], style={"width": "30%", "display": "inline-block", "verticalAlign": "top", "padding": "10px"}),
+
+    html.Div([
+        html.H4("Heatmap"),
+        dcc.Graph(id="heatmap-graph"),
+        html.H4("UMAP"),
+        dcc.Graph(id="umap-graph"),
+    ], style={"width": "65%", "display": "inline-block", "paddingLeft": "20px"})
+])
+
+# -------------------------
+# Robust callback
+# -------------------------
+@app.callback(
+    Output("heatmap-graph", "figure"),
+    Output("umap-graph", "figure"),
+    Input("heat-backend", "value"),
+    Input("umap-backend", "value"),
+    Input("normalizer", "value"),
+    Input("clusterer", "value"),
+    Input("cmap_scale", "value"),
+    Input("pca-components", "value")
+)
+def update_plots(heat_backend, umap_backend, normalizer, clusterer, cmap_scale, pca_components):
+    try:
+        # Always use plotly_express internally for UMAP in CombinedPolicy
+        combo = CombinedPolicy(
+            normalizer=normalizer,
+            heat_clusterer=clusterer,
+            heat_cmap_scale=cmap_scale,
+            heat_plotter=heat_backend,
+            heat_plot_kwargs=dict(show=False),
+            pre_umap_pca=pca_components if pca_components > 0 else None,
+            umap_scatter="plotly_express",
+            umap_scatter_kwargs=dict(show=False),
+        )
+
+        (heat_fig, umap_fig), meta = combo.apply(X, row_labels=rows, col_labels=cols, point_labels=point_labels)
+
+        # --- Heatmap backend ---
+        if heat_backend == "matplotlib":
+            heat_fig = mpl_to_dash_figure(plot_heat_mpl_dash(meta["X_clustered"], rows, cols))
+
+        # --- UMAP backend ---
+        if umap_backend == "matplotlib":
+            emb = meta.get("embedding")
+            if emb is None or emb.shape[1] < 2:
+                print("UMAP embedding invalid:", emb)
+                umap_fig = {"data": [], "layout": {"title":"UMAP embedding not available"}}
+            else:
+                umap_fig = mpl_to_dash_figure(plot_umap_mpl_dash(emb, point_labels))
+        # else keep Plotly figure as is
+
+        return heat_fig, umap_fig
+
+    except Exception as e:
+        print("Callback error:", e)
+        fallback_fig = {"data": [], "layout": {"title": "Error in figure"}}
+        return fallback_fig, fallback_fig
+
+# -------------------------
+if __name__ == "__main__":
+    import webbrowser
+    webbrowser.open("http://127.0.0.1:8050/")
+    app.run(debug=True)

src/visualization/plot.py

@@ -0,0 +1,214 @@
+# flexvis_app.py
+from __future__ import annotations
+
+# --- Matplotlib'i non-interactive moda al ---
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+
+import io, base64, inspect
+import numpy as np
+from typing import Optional, Dict, Any
+
+# --- Dash ---
+from dash import Dash, html, dcc, Input, Output
+
+# --- CombinedPolicy/REG: iki farklı modülden biri olabilir ---
+try:
+    from .flexvis import CombinedPolicy, REG  # varsa bunu kullan
+except ModuleNotFoundError:
+    from .flexvis import CombinedPolicy, REG              # yoksa bunu dene
+
+# =============================================================================
+# Adapters (MPL -> Dash image figure)
+# =============================================================================
+def fig_to_base64(fig) -> str:
+    """Matplotlib Figure -> base64 PNG (tarayıcıda gösterim için)"""
+    buf = io.BytesIO()
+    fig.savefig(buf, format="png", bbox_inches="tight")
+    plt.close(fig)
+    buf.seek(0)
+    return "data:image/png;base64," + base64.b64encode(buf.read()).decode("utf-8")
+
+def mpl_image_figure(img64: str) -> dict:
+    """Base64 PNG'i Dash Graph'ın kabul ettiği figure=dict yapısına sar."""
+    return {
+        "data": [],
+        "layout": {
+            "images": [{
+                "source": img64,
+                "xref": "paper",
+                "yref": "paper",
+                "x": 0, "y": 1,
+                "sizex": 1, "sizey": 1,
+                "xanchor": "left", "yanchor": "top",
+                "layer": "below"
+            }],
+            "xaxis": {"visible": False, "range": [0, 1]},
+            "yaxis": {"visible": False, "range": [0, 1], "scaleanchor": "x"},
+            "margin": {"l": 0, "r": 0, "t": 0, "b": 0},
+        }
+    }
+
+def to_dash_figure(fig_obj):
+    """
+    Plotly fig ise direkt döndür, Matplotlib fig ise base64'e çevirip imaj olarak döndür.
+    """
+    # Plotly fig kontrolü: obje 'to_dict' metoduna sahipse güvenle geçilebilir
+    if hasattr(fig_obj, "to_dict"):
+        return fig_obj
+    # Matplotlib Figure ise
+    return mpl_image_figure(fig_to_base64(fig_obj))
+
+# =============================================================================
+# CombinedPolicy kurulumunu güvenli yapan yardımcı
+# =============================================================================
+def _supported_params() -> set:
+    """CombinedPolicy'nin desteklediği __init__ parametrelerini oku."""
+    return set(inspect.signature(CombinedPolicy).parameters.keys())
+
+def build_combo(
+    heat_backend: str,
+    umap_backend: str,
+    normalizer: str,
+    clusterer: str,
+    cmap_scale: str,
+    pre_umap_pca: Optional[int],
+) -> CombinedPolicy:
+    """
+    CombinedPolicy'yi kurar. CombinedPolicy imzasında bulunmayan anahtarları
+    otomatik olarak filtreler (sürüm farklarına dayanıklı).
+    """
+    sup = _supported_params()
+
+    # UMAP scatter backend tercih sırası
+    if umap_backend in REG.scatter_plotters:
+        umap_scatter = umap_backend
+    elif "plotly_express" in REG.scatter_plotters:
+        umap_scatter = "plotly_express"
+    elif "plotly" in REG.scatter_plotters:
+        umap_scatter = "plotly"
+    else:
+        umap_scatter = "matplotlib"
+
+    # Iskelet konfig
+    cfg: Dict[str, Any] = {
+        # ortak
+        "normalizer": normalizer,
+        # heatmap
+        "heat_clusterer": clusterer,
+        "heat_cmap_scale": cmap_scale,
+        "heat_plotter": heat_backend,
+        "heat_plot_kwargs": dict(show=False),
+        # umap
+        "umap_neighbors": "sklearn",
+        "umap_reducer": "umap",
+        "umap_scatter": umap_scatter,
+        "umap_neighbors_kwargs": dict(n_neighbors=15, metric="euclidean"),
+        "umap_reducer_kwargs": dict(n_components=2, n_neighbors=15, min_dist=0.1,
+                                    metric="euclidean", random_state=7),
+        "umap_scatter_kwargs": dict(show=False),
+        # opsiyonel: bazı sürümlerde yok
+        "pre_umap_pca": pre_umap_pca,
+    }
+
+    # Sadece CombinedPolicy'nin bildiği anahtarları geçir
+    safe_cfg = {k: v for k, v in cfg.items() if (k in sup and v is not None)}
+    return CombinedPolicy(**safe_cfg)
+
+# =============================================================================
+# Demo veri (yerine kendi verini koyabilirsin)
+# =============================================================================
+rng = np.random.default_rng(42)
+X = rng.normal(size=(40, 25)) * 2 + rng.normal(size=(40, 1)) * 3
+rows = [f"Gene_{i}" for i in range(X.shape[0])]
+cols = [f"Cell_{j}" for j in range(X.shape[1])]
+point_labels = [f"C{i}" for i in range(X.shape[0])]
+
+# =============================================================================
+# Dash Uygulaması
+# =============================================================================
+app = Dash(__name__)
+app.layout = html.Div([
+    html.Div([
+        html.Label("Heatmap backend"),
+        dcc.Dropdown(["matplotlib", "plotly"], "plotly", id="heat-backend"),
+
+        html.Label("UMAP backend"),
+        dcc.Dropdown(["plotly_express", "matplotlib"], "plotly_express", id="umap-backend"),
+
+        html.Label("Normalizer"),
+        dcc.Dropdown(list(REG.normalizers.keys()), "zscore_rows", id="normalizer"),
+
+        html.Label("Clustering"),
+        dcc.Dropdown(list(REG.clusterers.keys()), "simple", id="clusterer"),
+
+        html.Label("Colormap scale"),
+        dcc.Dropdown(list(REG.cmap_scales.keys()), "center_zero", id="cmap_scale"),
+
+        html.Label("Pre-UMAP PCA components"),
+        dcc.Slider(min=0, max=X.shape[1], step=1, value=min(10, X.shape[1]),
+                   marks={i: str(i) for i in range(0, X.shape[1] + 1)}, id="pca-components"),
+    ], style={"width": "30%", "display": "inline-block", "verticalAlign": "top", "padding": "10px"}),
+
+    html.Div([
+        html.H4("Heatmap"),
+        dcc.Graph(id="heatmap-graph"),
+
+        html.H4("UMAP"),
+        dcc.Graph(id="umap-graph"),
+    ], style={"width": "65%", "display": "inline-block", "paddingLeft": "20px"})
+])
+
+# =============================================================================
+# Callback
+# =============================================================================
+@app.callback(
+    Output("heatmap-graph", "figure"),
+    Output("umap-graph", "figure"),
+    Input("heat-backend", "value"),
+    Input("umap-backend", "value"),
+    Input("normalizer", "value"),
+    Input("clusterer", "value"),
+    Input("cmap_scale", "value"),
+    Input("pca-components", "value"),
+)
+def update_plots(heat_backend, umap_backend, normalizer, clusterer, cmap_scale, pca_components):
+    try:
+        pre_pca = int(pca_components) if (pca_components is not None and int(pca_components) > 0) else None
+
+        combo = build_combo(
+            heat_backend=heat_backend,
+            umap_backend=umap_backend,
+            normalizer=normalizer,
+            clusterer=clusterer,
+            cmap_scale=cmap_scale,
+            pre_umap_pca=pre_pca,
+        )
+
+        # Pipeline’i çalıştır
+        (heat_fig, umap_fig), meta = combo.apply(
+            X, row_labels=rows, col_labels=cols, point_labels=point_labels,
+            heat_title="Heatmap", umap_title="UMAP"
+        )
+
+        # Dash'e uygun figür nesneleri
+        heat_out = to_dash_figure(heat_fig)
+        umap_out = to_dash_figure(umap_fig)
+
+        return heat_out, umap_out
+
+    except Exception as e:
+        # Hata durumunda ekrana yaz ve boş figür dön
+        msg = f"Callback error: {type(e).__name__}: {e}"
+        print(msg)
+        fallback = {"data": [], "layout": {"title": msg}}
+        return fallback, fallback
+
+# =============================================================================
+# Main
+# =============================================================================
+if __name__ == "__main__":
+    # İstersen tarayıcıyı otomatik açabilirsin:
+    # import webbrowser; webbrowser.open("http://127.0.0.1:8050/")
+    app.run(debug=True)