makoview 0.1.2__py3-none-any.whl

makoview/app.py

@@ -0,0 +1,179 @@
+from shiny.types import SilentException
+from shiny.express import input, render, ui
+from shiny import reactive
+import duckdb
+import os
+import pandas as pd
+from pathlib import Path
+
+import shinyswatch
+
+import plots
+
+_db_path = os.environ["MAKO_DIFFERENTIAL_DB"]
+_reads_path = os.environ["MAKO_MODIFICATION_DB"]
+
+# Validate that database files exist
+if not os.path.exists(_db_path):
+    raise FileNotFoundError(f"Differential sites database not found: {_db_path}")
+if not os.path.exists(_reads_path):
+    raise FileNotFoundError(f"Modification database not found: {_reads_path}")
+
+
+# load from db into a Pandas dataframe in-memory for faster access
+def load_data_into_memory(db_path) -> pd.DataFrame:
+    conn = duckdb.connect(database=db_path, read_only=True)
+    df = conn.execute("SELECT * FROM sites").fetchdf()
+    conn.close()
+    return df
+
+
+df = load_data_into_memory(_db_path)
+
+
+# UI Definition
+ui.page_opts(title="makoview", theme=shinyswatch.theme.lumen)
+ui.include_css(Path(__file__).parent / "styles.css")
+# ui.page_opts(title="Mako modification lookup", fillable=True)
+
+read_cache: reactive.Value[pd.DataFrame] = reactive.value()
+
+
+with ui.sidebar(width="400px", open="always"):
+    with ui.card(fill=False):
+        ui.card_header("Search for a transcript...")
+        # with ui.layout_columns(col_widths=(8, 4), gap="0.75rem", row_heights="auto"):
+        ui.input_text(
+            "transcript_id",
+            "Transcript ID",
+            placeholder="e.g. ENST00000000233.10",
+            width="100%",
+        )
+
+        ui.input_selectize(
+            "transcript_position",
+            "Transcript Position",
+            choices=[],
+            multiple=False,
+            options={
+                "placeholder": "transcript not found...",
+                "dropdownParent": "body",
+            },
+            width="100%",
+        )
+        ui.input_action_button(
+            "search_btn", "Search", class_="btn-primary", style="margin-top: 0.5rem;"
+        )
+
+    @render.data_frame
+    @reactive.event(input.search_btn, ignore_none=False)
+    def on_search():
+        """Generate boxplot of probability_modified by group and sample."""
+        transcript_id: str = input.transcript_id()
+        transcript_position: int = input.transcript_position()
+
+        if not transcript_id or not transcript_id.strip():
+            return None
+
+        # Get matching reads as a dataframe
+        reads_df = get_matching_reads(transcript_id.strip(), int(transcript_position))
+
+        site_df = df.loc[
+            (df["transcript_id"] == transcript_id.strip())
+            & (df["transcript_position"] == int(transcript_position))
+        ]
+        site_df_tidy = site_df.melt(var_name="column", value_name="value")
+
+        read_cache.set(reads_df)
+
+        return render.DataGrid(data=site_df_tidy, width="100%", height="auto")
+
+
+def get_matching_reads(transcript_id: str, transcript_position: int) -> pd.DataFrame:
+    """
+    Find rnames for the given transcript_id and transcript_position,
+    then query reads database for matching non-ignored reads.
+    """
+    # Find matching rnames from the sites dataframe
+    matching_rnames = (
+        df.loc[
+            (df["transcript_id"] == transcript_id)
+            & (df["transcript_position"] == transcript_position),
+            "rname",
+        ]
+        .unique()
+        .tolist()
+    )
+
+    print(matching_rnames)
+
+    if not matching_rnames:
+        return pd.DataFrame()
+
+    # Query reads database for matching, non-ignored reads
+    conn = duckdb.connect(database=_reads_path, read_only=True)
+
+    # Use parameterized query to avoid SQL injection
+    placeholders = ", ".join(["?" for _ in matching_rnames])
+    query = f"""
+        SELECT * FROM reads
+        WHERE rname IN ({placeholders})
+          AND transcript_position = {transcript_position}
+          AND ignored = FALSE
+    """
+
+    reads_df = conn.execute(query, matching_rnames).fetchdf()
+    conn.close()
+
+    return reads_df
+
+
+# Server logic
+@reactive.effect
+@reactive.event(input.transcript_id)
+def update_transcript_pos():
+    """Update the transcript position options based on the entered transcript ID."""
+    transcript_id: str = input.transcript_id()
+
+    if transcript_id and transcript_id.strip():
+        positions = df.loc[
+            df["transcript_id"] == transcript_id.strip(), "transcript_position"
+        ].unique()
+        positions = sorted(positions.tolist())
+    else:
+        positions = []
+
+    ui.update_selectize(
+        "transcript_position",
+        choices=positions,
+    )
+
+
+@render.data_frame
+def plot_counts():
+    try:
+        subset = read_cache.get()
+    except SilentException:
+        return None
+
+    binarized_df = (
+        subset.groupby(["sample_name", "group_name"])
+        .agg(
+            successes=("probability_modified", lambda x: (x >= 0.5).sum()),
+            failures=("probability_modified", lambda x: (x < 0.5).sum()),
+        )
+        .sort_values(["group_name", "sample_name"])
+        .reset_index()
+    )
+
+    return binarized_df
+
+
+@render.plot(height=750)
+def plot_modification():
+    try:
+        subset = read_cache.get()
+    except SilentException:
+        return None
+
+    return plots.plot_binarised_violin_by_site(subset)

makoview/plots.py

@@ -0,0 +1,105 @@
+import pandas as pd
+import matplotlib.pyplot as plt
+import seaborn as sns
+import numpy as np
+
+
+def plot_binarised_violin_by_site(subset):
+    # Binarise probability (threshold 0.5)
+    subset["binarised_probability_modified"] = (
+        subset["probability_modified"] >= 0.5
+    ).astype(int)
+
+    # Add combined label and sort
+    subset["label"] = subset["sample_name"] + " (" + subset["group_name"] + ")"
+    subset.sort_values(["group_name", "sample_name"], inplace=True)
+    label_order = subset["label"].unique()
+
+    # Compute counts per label
+    counts = subset.groupby("label")["binarised_probability_modified"]
+    n = counts.size().reindex(label_order)
+    T = counts.sum().reindex(label_order)
+    F = n - T
+    T_over_n = (T / n).round(3)
+
+    label_order_with_stats = [
+        f"{lbl}\n(n={n[lbl]}, T={T[lbl]}, F={F[lbl]}, T/n={T_over_n[lbl]})"
+        for lbl in label_order
+    ]
+
+    # Plot
+    fig, ax = plt.subplots(1, 1, figsize=(max(10, len(label_order) * 1.5), 8))
+
+    sns.violinplot(
+        data=subset,
+        x="label",
+        y="binarised_probability_modified",
+        order=label_order,
+        inner=None,
+        density_norm="width",
+        cut=0,
+        color="skyblue",
+        ax=ax,
+    )
+
+    sns.boxplot(
+        data=subset,
+        x="label",
+        y="binarised_probability_modified",
+        order=label_order,
+        showcaps=True,
+        width=0.15,
+        boxprops={"facecolor": "white", "edgecolor": "black", "linewidth": 1},
+        whiskerprops={"color": "black", "linewidth": 1},
+        capprops={"color": "black", "linewidth": 1},
+        medianprops={"color": "black", "linewidth": 1},
+        showfliers=False,
+        ax=ax,
+    )
+
+    sns.stripplot(
+        data=subset,
+        x="label",
+        y="binarised_probability_modified",
+        order=label_order,
+        color="black",
+        size=3,
+        jitter=True,
+        alpha=0.5,
+        ax=ax,
+    )
+
+    # Mean ± SD
+    stats = (
+        subset.groupby("label")["binarised_probability_modified"]
+        .agg(["mean", "std"])
+        .reindex(label_order)
+    )
+
+    for j, label in enumerate(label_order):
+        mean_val = stats.loc[label, "mean"]
+        std_val = stats.loc[label, "std"]
+
+        ax.plot(j, mean_val, "o", color="red", markersize=6)
+
+        ymin = max(0, mean_val - std_val)
+        ymax = min(1, mean_val + std_val)
+        ax.errorbar(
+            j,
+            mean_val,
+            yerr=[[mean_val - ymin], [ymax - mean_val]],
+            fmt="none",
+            ecolor="red",
+            elinewidth=1,
+            capsize=5,
+        )
+
+    ax.set_xticks(range(len(label_order)))
+    ax.set_xticklabels(label_order_with_stats, rotation=45, ha="right")
+    ax.set_xlabel("Sample (Group)")
+    ax.set_ylabel("Binarised Probability Modified")
+    ax.set_title("Violin Plot of Binarised Probability Modified for each Sample")
+    ax.set_ylim(-0.05, 1.05)
+    ax.grid(True, linestyle="--", alpha=0.6)
+
+    return fig

makoview/styles.css

@@ -0,0 +1,3 @@
+.navbar {
+    border: none;
+}

makoview/wrapper.py

@@ -0,0 +1,63 @@
+from shiny import run_app
+import os
+import argparse
+from pathlib import Path
+
+
+def main():
+    """Launch the MakoView Shiny visualization application."""
+
+    parser = argparse.ArgumentParser(
+        description="Launch makoview",
+    )
+
+    parser.add_argument(
+        "--differential-results",
+        type=Path,
+        required=True,
+        dest="differential_results",
+        help="Path to differential sites database file",
+    )
+
+    parser.add_argument(
+        "--modification-db",
+        type=Path,
+        required=True,
+        dest="modification_db",
+        help="Path to modification database file",
+    )
+
+    parser.add_argument(
+        "--port",
+        type=int,
+        default=8000,
+        help="Port for the Shiny application (default: 8000)",
+    )
+
+    args = parser.parse_args()
+
+    # Use explicitly provided paths
+    diff_db_path = args.differential_results
+    mod_db_path = args.modification_db
+
+    # Validate files exist
+    if not diff_db_path.exists():
+        parser.error(f"Differential results file not found: {diff_db_path}")
+    if not mod_db_path.exists():
+        parser.error(f"Modification database file not found: {mod_db_path}")
+
+    # Set environment variables for app.py to consume
+    os.environ["MAKO_DIFFERENTIAL_DB"] = str(diff_db_path.absolute())
+    os.environ["MAKO_MODIFICATION_DB"] = str(mod_db_path.absolute())
+
+    print(f"Starting Mako Shiny app on port {args.port}...")
+    print(f"  Differential DB: {diff_db_path.absolute()}")
+    print(f"  Modification DB: {mod_db_path.absolute()}")
+
+    path = Path(__file__).parent.resolve()
+
+    run_app(str(path / "app.py"), port=args.port)  # type: ignore[call-non-callable]
+
+
+if __name__ == "__main__":
+    main()

makoview-0.1.2.dist-info/METADATA

@@ -0,0 +1,14 @@
+Metadata-Version: 2.4
+Name: makoview
+Version: 0.1.2
+Summary: Visualise the results from Mako
+Requires-Python: <3.13,>=3.9
+Requires-Dist: duckdb>=1.0.0
+Requires-Dist: matplotlib>=3.9.4
+Requires-Dist: pandas>=2.3.3
+Requires-Dist: seaborn>=0.13.2
+Requires-Dist: shiny>=0.10.0
+Requires-Dist: shinyswatch>=0.9.0
+Description-Content-Type: text/markdown
+
+# makoview

makoview-0.1.2.dist-info/RECORD

@@ -0,0 +1,9 @@
+makoview/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+makoview/app.py,sha256=M_jU2T-c4PpSfSyIqqV5ih3w3k46SQbIkJ264I6xPLs,5234
+makoview/plots.py,sha256=yUQ3BPZmhJPJPF4LCT1qnyiqLSOiTh62dPifYbfFQgU,2970
+makoview/styles.css,sha256=ttB0IB8J7_nxmwmddHoEIOvDB2QaWNoIdIS3OVq80Qo,29
+makoview/wrapper.py,sha256=xaOdKwvobRLs81xktfWTbzZX2JVzwN9yksGV2etISRY,1724
+makoview-0.1.2.dist-info/METADATA,sha256=HxT0Hx08lJsO-njGJxM0_C-1KzGt5_LLIkYmy5JCvc4,358
+makoview-0.1.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+makoview-0.1.2.dist-info/entry_points.txt,sha256=ezLJK-IIC64DKimCa4fYFeVEaxHF1l0QFxwC-aL25IM,51
+makoview-0.1.2.dist-info/RECORD,,

makoview-0.1.2.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: hatchling 1.28.0
+Root-Is-Purelib: true
+Tag: py3-none-any

makoview-0.1.2.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+makoview = makoview.wrapper:main