histoseg 0.1.8.1__py3-none-any.whl → 0.1.9__py3-none-any.whl

histoseg/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.1.8.1'
-__version_tuple__ = version_tuple = (0, 1, 8, 1)
+__version__ = version = '0.1.9'
+__version_tuple__ = version_tuple = (0, 1, 9)
 
 __commit_id__ = commit_id = None

histoseg/contours/pattern1_isoline.py

@@ -40,14 +40,47 @@ from sklearn.neighbors import KNeighborsRegressor
 PathLike = Union[str, Path]
 
 
+def _normalize_cluster_label(x: object) -> str:
+    """Normalize cluster label to a stable string.
+
+    Goals:
+      - Support numeric clusters: 10, 10.0, "10", "10.0" -> "10"
+      - Support text clusters: "Luminal A" stays "Luminal A" (trim whitespace)
+      - Treat NaN/None as "" (caller should drop or error)
+    """
+    if x is None:
+        return ""
+    # pandas NA
+    try:
+        if pd.isna(x):  # type: ignore[arg-type]
+            return ""
+    except Exception:
+        pass
+
+    s = str(x).strip()
+    if s == "":
+        return ""
+    # "10.0" -> "10" (only if it is exactly an integer-looking float string)
+    # This avoids mangling genuine text like "A.0".
+    if s.endswith(".0"):
+        head = s[:-2]
+        if head.isdigit():
+            return head
+    return s
+
+
 @dataclass(frozen=True)
 class Pattern1IsolineConfig:
-    # Required inputs
+    # Required inputs (non-defaults MUST come first for dataclasses)
     clusters_csv: PathLike
     cells_parquet: PathLike
-    tissue_boundary_csv: Optional[PathLike]
     out_dir: PathLike
-    pattern1_clusters: Sequence[int]
+    pattern1_clusters: Sequence[Union[int, str]]
+
+    # Optional inputs / schema controls
+    tissue_boundary_csv: Optional[PathLike] = None
+    barcode_col: str = "Barcode"
+    cluster_col: str = "Cluster"
 
     # Core params (defaults match the notebook)
     grid_n: int = 1200
@@ -74,6 +107,16 @@ class Pattern1IsolineConfig:
     # Contour
     isoline_level: float = 0.5
 
+    # Labeling scheme
+    #   - "p1_is_one":   pattern1=1, others=0 (original behavior)
+    #   - "p1_is_zero":  pattern1=0, others=1 (requested alternative)
+    label_scheme: str = "p1_is_one"
+
+    # Segmentation confidence score (cophenetic blue-band mean)
+    compute_confidence_score: bool = False
+    confidence_linkage_method: str = "average"
+    confidence_show_corr: bool = False
+
     # Output controls
     save_params_json: bool = True
     save_contours_npy: bool = True
@@ -92,10 +135,21 @@ class Pattern1IsolineResult:
     n_target_cells: int
     n_bg0_points: int
     contours: List[np.ndarray]
+    label_scheme: str
+    segmentation_confidence_score: Optional[float] = None
+    segmentation_confidence_stats: Optional[Mapping[str, Union[int, float]]] = None
     params_json: Optional[Path] = None
     preview_png: Optional[Path] = None
 
 
+@dataclass
+class SegmentationConfidenceResult:
+    """Result container for the cophenetic blue-band score."""
+    score_mean: float
+    stats: Mapping[str, Union[int, float]]
+    blue_band_matrix: Optional[pd.DataFrame] = None
+
+
 def _make_jupyterlab_tree_href(path: Path) -> str:
     """Build an href that opens `path` in JupyterLab's file browser.
 
@@ -305,6 +359,8 @@ def sample_background_from_other_cells_plus_synth(
 def align_clusters_with_cells(
     clusters_csv: PathLike,
     cells_parquet: PathLike,
+    barcode_col: str = "Barcode",
+    cluster_col: str = "Cluster",
 ) -> Tuple[pd.DataFrame, str, str, str]:
     """Align clusters.csv(Barcode/Cluster) with cells.parquet.
 
@@ -314,18 +370,25 @@ def align_clusters_with_cells(
         x_col, y_col: chosen coordinate columns
     """
     cl = pd.read_csv(clusters_csv)
-    if "Barcode" not in cl.columns or "Cluster" not in cl.columns:
-        raise ValueError(f"clusters.csv 需要包含 Barcode/Cluster 列，当前列={list(cl.columns)}")
+    if barcode_col not in cl.columns or cluster_col not in cl.columns:
+        raise ValueError(
+            f"clusters.csv 需要包含 {barcode_col}/{cluster_col} 列，"
+            f"当前列={list(cl.columns)}"
+        )
 
     cl = cl.copy()
-    cl["Barcode"] = cl["Barcode"].astype(str)
-    cl["Cluster"] = pd.to_numeric(cl["Cluster"], errors="coerce").astype("Int64")
+    cl[barcode_col] = cl[barcode_col].astype(str).str.strip()
+    # IMPORTANT: keep cluster labels as string to support non-numeric clusters.
+    cl[cluster_col] = cl[cluster_col].map(_normalize_cluster_label)
+
+    # Drop rows with empty barcode or empty cluster
+    cl = cl.loc[(cl[barcode_col] != "") & (cl[cluster_col] != ""), [barcode_col, cluster_col]].copy()
 
     cells = pd.read_parquet(cells_parquet)
 
     # Try to infer coordinate columns
-    cand_x = [c for c in cells.columns if c.lower() in ["x", "x_centroid", "x_center", "xcoord", "x_coord"]]
-    cand_y = [c for c in cells.columns if c.lower() in ["y", "y_centroid", "y_center", "ycoord", "y_coord"]]
+    cand_x = [c for c in cells.columns if c.lower() in ["x", "x_centroid", "x_center", "xcoord", "x_coord", "x_centroid_um", "x_centroid_px", "x_centroid_microns"]]
+    cand_y = [c for c in cells.columns if c.lower() in ["y", "y_centroid", "y_center", "ycoord", "y_coord", "y_centroid_um", "y_centroid_px", "y_centroid_microns"]]
     if not cand_x or not cand_y:
         raise ValueError(f"cells.parquet 找不到 x/y 列。列名示例：{list(cells.columns)[:60]}")
 
@@ -349,13 +412,13 @@ def align_clusters_with_cells(
         tmp = cells.copy()
         tmp["_join_id"] = tmp[cells_id_col].astype(str)
         cl2 = cl.copy()
-        cl2["_join_id"] = cl2["Barcode"].astype(str)
+        cl2["_join_id"] = cl2[barcode_col].astype(str)
 
         if strip_suffix:
             tmp["_join_id"] = tmp["_join_id"].str.replace(r"-1$", "", regex=True)
             cl2["_join_id"] = cl2["_join_id"].str.replace(r"-1$", "", regex=True)
 
-        m = tmp.merge(cl2[["_join_id", "Cluster"]], on="_join_id", how="inner")
+        m = tmp.merge(cl2[["_join_id", cluster_col]], on="_join_id", how="inner")
         return m
 
     best: Optional[pd.DataFrame] = None
@@ -372,32 +435,190 @@ def align_clusters_with_cells(
         # Provide debugging hints
         msg = [
             "[FAIL] 无法将 clusters.csv 的 Barcode 对齐到 cells.parquet",
-            f"clusters.csv Barcode 示例: {cl['Barcode'].head().tolist()}",
+            f"clusters.csv {barcode_col} 示例: {cl[barcode_col].head().tolist()}",
             f"cells.parquet 列名: {list(cells.columns)[:80]}",
         ]
         for c in id_candidates[:6]:
             msg.append(f"cells[{c}] 示例: {cells[c].astype(str).head().tolist()}")
         raise RuntimeError("\n".join(msg))
 
-    id_col_used, stripped = best_info
-    # Rename Cluster -> cluster (int)
-    out = best.rename(columns={"Cluster": "cluster"})
+    id_col_used, _stripped = best_info
+    # Rename cluster_col -> cluster (string label)
+    out = best.rename(columns={cluster_col: "cluster"})
+    out["cluster"] = out["cluster"].map(_normalize_cluster_label)
     return out, id_col_used, x_col, y_col
 
 
+def _validate_label_scheme(label_scheme: str) -> str:
+    s = str(label_scheme).strip().lower()
+    if s in {"p1_is_one", "pattern1_is_one", "p1=1", "one"}:
+        return "p1_is_one"
+    if s in {"p1_is_zero", "pattern1_is_zero", "p1=0", "zero", "invert", "inverted"}:
+        return "p1_is_zero"
+    raise ValueError(
+        f"Unknown label_scheme={label_scheme!r}. Supported: 'p1_is_one' or 'p1_is_zero'."
+    )
+
+
+def compute_segmentation_confidence_score_from_merged(
+    merged_df: pd.DataFrame,
+    *,
+    pattern1_clusters: Sequence[Union[int, str]],
+    x_col: str,
+    y_col: str,
+    celltype_col: str = "cluster",
+    z_col: Optional[str] = None,
+    linkage_method: str = "average",
+    show_corr: bool = False,
+    return_blue_band_matrix: bool = False,
+) -> SegmentationConfidenceResult:
+    """Compute the segmentation confidence score (cophenetic blue-band mean).
+
+    Definition (as requested):
+      - Build the searcher→findee distance matrix at the cluster level.
+      - Compute the cophenetic (ultrametric) matrix from hierarchical clustering.
+      - Take all cophenetic values between PATTERN1 clusters (A) and non-PATTERN1 clusters (B).
+      - Return the mean of that cross-block ("blue band") as the confidence score.
+
+    Notes:
+      - This function is intentionally *separate* from isoline generation so you can call it independently.
+      - It uses histoseg.sfplot.Searcher_Findee_Score utilities.
+
+    Returns:
+      SegmentationConfidenceResult(score_mean, stats, blue_band_matrix?)
+    """
+    required = {x_col, y_col, celltype_col}
+    missing = required - set(merged_df.columns)
+    if missing:
+        raise ValueError(f"merged_df missing required columns: {sorted(missing)}")
+
+    # Lazy import: keep this module importable even if sfplot deps are optional.
+    from histoseg.sfplot.Searcher_Findee_Score import (
+        compute_searcher_findee_distance_matrix_from_df,
+        compute_cophenetic_from_distance_matrix,
+    )
+
+    df = merged_df.copy()
+    df[celltype_col] = df[celltype_col].map(_normalize_cluster_label)
+    df = df.loc[df[celltype_col] != "", [x_col, y_col, celltype_col] + ([z_col] if z_col else [])].copy()
+
+    if df[celltype_col].nunique() < 2:
+        raise ValueError("Need at least 2 clusters to compute cophenetic score.")
+
+    distance_matrix = compute_searcher_findee_distance_matrix_from_df(
+        df,
+        x_col=x_col,
+        y_col=y_col,
+        z_col=z_col,
+        celltype_col=celltype_col,
+    )
+
+    if getattr(distance_matrix, "shape", (0, 0))[0] < 2:
+        raise ValueError("distance_matrix too small; check cluster sizes.")
+
+    row_coph, col_coph = compute_cophenetic_from_distance_matrix(
+        distance_matrix,
+        method=linkage_method,
+        show_corr=show_corr,
+    )
+
+    # Use row_coph by default (matches the user's notebook code)
+    coph = row_coph
+
+    labels = pd.Index(coph.index).map(_normalize_cluster_label)
+    coph = coph.copy()
+    coph.index = labels
+    coph.columns = labels
+
+    A = pd.Index([_normalize_cluster_label(x) for x in pattern1_clusters if _normalize_cluster_label(x) != ""]).unique()
+    A = A.intersection(coph.index)
+
+    if len(A) == 0:
+        raise ValueError("No PATTERN1 clusters found in cophenetic matrix index. Check labels.")
+    B = pd.Index(coph.index).difference(A)
+    if len(B) == 0:
+        raise ValueError("No non-PATTERN1 clusters found; cannot compute cross-block mean.")
+
+    blue_band_matrix = coph.loc[A, B]
+    band = blue_band_matrix.to_numpy().ravel()
+    band = band[~np.isnan(band)]
+
+    if band.size == 0:
+        raise ValueError("blue band has no finite values; cannot compute score.")
+
+    stats: Mapping[str, Union[int, float]] = {
+        "n_pairs": int(band.size),
+        "min": float(np.min(band)),
+        "p05": float(np.quantile(band, 0.05)),
+        "median": float(np.median(band)),
+        "mean": float(np.mean(band)),
+        "p95": float(np.quantile(band, 0.95)),
+        "max": float(np.max(band)),
+    }
+
+    return SegmentationConfidenceResult(
+        score_mean=float(stats["mean"]),
+        stats=stats,
+        blue_band_matrix=blue_band_matrix if return_blue_band_matrix else None,
+    )
+
+
+def compute_segmentation_confidence_score(
+    *,
+    clusters_csv: PathLike,
+    cells_parquet: PathLike,
+    pattern1_clusters: Sequence[Union[int, str]],
+    barcode_col: str = "Barcode",
+    cluster_col: str = "Cluster",
+    linkage_method: str = "average",
+    show_corr: bool = False,
+    return_blue_band_matrix: bool = False,
+) -> SegmentationConfidenceResult:
+    """Convenience wrapper: load files, merge, and compute confidence score."""
+    merged, _id_col_used, x_col, y_col = align_clusters_with_cells(
+        clusters_csv=clusters_csv,
+        cells_parquet=cells_parquet,
+        barcode_col=barcode_col,
+        cluster_col=cluster_col,
+    )
+    return compute_segmentation_confidence_score_from_merged(
+        merged,
+        pattern1_clusters=pattern1_clusters,
+        x_col=x_col,
+        y_col=y_col,
+        celltype_col="cluster",
+        z_col=None,
+        linkage_method=linkage_method,
+        show_corr=show_corr,
+        return_blue_band_matrix=return_blue_band_matrix,
+    )
+
+
 def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
     """Run the full pipeline and (optionally) save outputs."""
     out_dir = Path(cfg.out_dir)
     out_dir.mkdir(parents=True, exist_ok=True)
 
-    merged, id_col_used, x_col, y_col = align_clusters_with_cells(cfg.clusters_csv, cfg.cells_parquet)
+    label_scheme = _validate_label_scheme(cfg.label_scheme)
+
+    merged, id_col_used, x_col, y_col = align_clusters_with_cells(
+        cfg.clusters_csv,
+        cfg.cells_parquet,
+        barcode_col=cfg.barcode_col,
+        cluster_col=cfg.cluster_col,
+    )
 
     merged = merged.copy()
-    merged["cluster"] = pd.to_numeric(merged["cluster"], errors="coerce").astype("Int64")
-    merged = merged.dropna(subset=["cluster"]).copy()
-    merged["cluster"] = merged["cluster"].astype(int)
+    merged["cluster"] = merged["cluster"].map(_normalize_cluster_label)
+    merged = merged.loc[merged["cluster"] != ""].copy()
+
+    # pattern1 cluster labels
+    p1 = set(_normalize_cluster_label(x) for x in cfg.pattern1_clusters)
+    p1 = {x for x in p1 if x != ""}
+
+    if len(p1) == 0:
+        raise ValueError("pattern1_clusters is empty after normalization.")
 
-    p1 = set(int(x) for x in cfg.pattern1_clusters)
     merged["_is_p1"] = merged["cluster"].isin(p1)
 
     p1_df = merged.loc[merged["_is_p1"], [id_col_used, x_col, y_col]].copy()
@@ -439,15 +660,25 @@ def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
     if len(bg0_xy) == 0:
         raise RuntimeError("No bg0 points sampled. Try relaxing bg_d_min/bg_d_max, or disabling synth bg.")
 
-    # Train KNN regressor: target=1, bg=0
+    # Train KNN regressor
+    #  - label_scheme="p1_is_one":  target=1, bg=0  (original)
+    #  - label_scheme="p1_is_zero": target=0, bg=1  (inverted)
     X_train = np.vstack([bg0_xy, target_xy])
-    y_train = np.hstack([np.zeros(len(bg0_xy)), np.ones(len(target_xy))])
+    if label_scheme == "p1_is_one":
+        y_train = np.hstack([np.zeros(len(bg0_xy)), np.ones(len(target_xy))])
+    else:
+        y_train = np.hstack([np.ones(len(bg0_xy)), np.zeros(len(target_xy))])
 
     reg = KNeighborsRegressor(n_neighbors=cfg.knn_k, weights="distance")
     reg.fit(X_train, y_train)
 
     # Predict on mesh + smooth
-    xx, yy, grid = make_mesh_from_xy(target_xy, grid_n=cfg.grid_n, pad_fraction=cfg.pad_fraction, margin_um=cfg.margin_um)
+    xx, yy, grid = make_mesh_from_xy(
+        target_xy,
+        grid_n=cfg.grid_n,
+        pad_fraction=cfg.pad_fraction,
+        margin_um=cfg.margin_um,
+    )
     prob = reg.predict(grid).reshape(xx.shape)
     prob_smooth = gaussian_filter(prob, sigma=cfg.smooth_sigma)
 
@@ -458,7 +689,7 @@ def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
     prob_smooth_masked = prob_smooth.copy()
     prob_smooth_masked[~tissue_mask] = np.nan
 
-    # 0.5 isoline
+    # 0.5 isoline (or cfg.isoline_level)
     verts_list = extract_contour_paths(xx, yy, prob_smooth_masked, level=cfg.isoline_level)
     verts_list = filter_loops_by_cell_count(verts_list, target_xy, min_cells_inside=cfg.min_cells_inside)
 
@@ -468,6 +699,24 @@ def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
             "建议：min_cells_inside 降低（如 10->3），smooth_sigma 增大（如 5->8），knn_k 增大（如 30->50），或降低 grid_n。"
         )
 
+    # Optional: compute segmentation confidence score (cophenetic blue-band mean)
+    conf_score: Optional[float] = None
+    conf_stats: Optional[Mapping[str, Union[int, float]]] = None
+    if cfg.compute_confidence_score:
+        conf_res = compute_segmentation_confidence_score_from_merged(
+            merged,
+            pattern1_clusters=cfg.pattern1_clusters,
+            x_col=x_col,
+            y_col=y_col,
+            celltype_col="cluster",
+            z_col=None,
+            linkage_method=cfg.confidence_linkage_method,
+            show_corr=cfg.confidence_show_corr,
+            return_blue_band_matrix=False,
+        )
+        conf_score = conf_res.score_mean
+        conf_stats = conf_res.stats
+
     params_path: Optional[Path] = None
     if cfg.save_params_json:
         params = asdict(cfg)
@@ -479,6 +728,9 @@ def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
                 n_target_cells=int(len(target_xy)),
                 n_bg0=int(len(bg0_xy)),
                 n_contours=int(len(verts_list)),
+                label_scheme=label_scheme,
+                segmentation_confidence_score=conf_score,
+                segmentation_confidence_stats=conf_stats,
             )
         )
         params_path = out_dir / "params.json"
@@ -497,7 +749,12 @@ def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
         for v in verts_list:
             plt.plot(v[:, 0], v[:, 1], linewidth=2)
         plt.gca().set_aspect("equal")
-        plt.title(f"Pattern1 segmentation | isoline={cfg.isoline_level:g} | contours={len(verts_list)}")
+
+        title = f"Pattern1 segmentation | isoline={cfg.isoline_level:g} | contours={len(verts_list)} | label_scheme={label_scheme}"
+        if conf_score is not None:
+            title += f" | confidence(mean)={conf_score:.4f}"
+        plt.title(title)
+
         plt.legend(frameon=False)
         plt.tight_layout()
 
@@ -528,6 +785,9 @@ def run_pattern1_isoline(cfg: Pattern1IsolineConfig) -> Pattern1IsolineResult:
         n_target_cells=int(len(target_xy)),
         n_bg0_points=int(len(bg0_xy)),
         contours=list(verts_list),
+        label_scheme=label_scheme,
+        segmentation_confidence_score=conf_score,
+        segmentation_confidence_stats=conf_stats,
         params_json=params_path,
         preview_png=preview_path,
     )
@@ -539,7 +799,7 @@ def run_pattern1_isoline_from_hf(
     *,
     revision: str = "main",
     out_dir: PathLike = "outputs/pattern1_isoline0p5_from_graphclust",
-    pattern1_clusters: Sequence[int] = (10, 23, 19, 27, 14, 20, 25, 26),
+    pattern1_clusters: Sequence[Union[int, str]] = (10, 23, 19, 27, 14, 20, 25, 26),
     clusters_relpath: str = "analysis/clustering/gene_expression_graphclust/clusters.csv",
     cache_dir: Optional[PathLike] = None,
     **cfg_overrides,

{histoseg-0.1.8.1.dist-info → histoseg-0.1.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: histoseg
-Version: 0.1.8.1
+Version: 0.1.9
 Summary: ...
 Author-email: Taobo Hu <taobo.hu@scilifelab.se>
 License: Required Notice: Copyright (c) 2025 SPATHO AB.
@@ -86,6 +86,14 @@ Project-URL: Documentation, https://histoseg.readthedocs.io
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: numpy
+Requires-Dist: pandas
+Requires-Dist: pyarrow
+Requires-Dist: scipy
+Requires-Dist: scikit-learn
+Requires-Dist: seaborn
+Requires-Dist: shapely
+Requires-Dist: matplotlib
 Dynamic: license-file
 
 <div align="center">
@@ -155,12 +163,6 @@ The Pattern1 isoline workflow uses:
 - matplotlib
 - a Parquet engine (**pyarrow is recommended**)
 
-If you run into missing imports, install them explicitly:
-
-```bash
-pip install -U numpy pandas pyarrow scipy scikit-learn matplotlib
-```
-
 Optional:
 
 - Hugging Face downloader: `pip install -U huggingface_hub`

{histoseg-0.1.8.1.dist-info → histoseg-0.1.9.dist-info}/RECORD

@@ -1,7 +1,7 @@
 histoseg/__init__.py,sha256=ssVwpgspXb7TWi1aAbWYA82FBJhGZH1g-JIvosVocTM,982
-histoseg/_version.py,sha256=DceX3WwB1oE06QTAMXFaSlGTc9twNCVWkTErPbrlF7o,709
+histoseg/_version.py,sha256=ib8ckvf-NNDfacXd8unW0p5cf-gl57XyQvjoEMc_pvc,704
 histoseg/contours/__init__.py,sha256=8YEy98MnGOhJg1BHkVZ_qPtTYUWPoIDAOT60F5EfHgY,32
-histoseg/contours/pattern1_isoline.py,sha256=6JVL26O7AZ0s0YyloDoiUdIHTJ7jp6gEooFWnMoNhSw,19592
+histoseg/contours/pattern1_isoline.py,sha256=ljEE-_Y0tWCRHTNytRDgWHgft6oqcFuwRQgZPwoMHTg,28836
 histoseg/geometry/__init__.py,sha256=5Aep5GBj2u4k2415QIECd0vK3y2eVBfQZRIskOgzb78,101
 histoseg/geometry/tissue_boundary.py,sha256=-12lyNYI93JkNlBDT_Fv7SLAeTLKMZ9fPhqpSTowFz0,3680
 histoseg/gui/__init__.py,sha256=SJvM-gRCwQ0X7fufPmdNO43X3wQvlnFHFmFNswt1hlw,133
@@ -9,8 +9,8 @@ histoseg/gui/gui_app.py,sha256=copFjJzMEeg2T4HV4--KHq9GG-v58xeW-eqRQ0-uJww,15259
 histoseg/io/__init__.py,sha256=kH_F15ApTutYbEUGAkV9QxBv8Ho863xvWK0mRmKeOCA,27
 histoseg/io/huggingface.py,sha256=YL_aXAXiHMMk6T6bqX-zPXFFChO83CVhTxI6mX52Z7g,2124
 histoseg/sfplot/Searcher_Findee_Score.py,sha256=Y9UzWrqap029BOyOnFPklvK4v-2wm3r_PkmUh65DEqo,14951
-histoseg-0.1.8.1.dist-info/licenses/LICENSE,sha256=z7Ztufk460DPfU3rgZEstjCQK3EbwHbD4JSmF_7y0qA,4764
-histoseg-0.1.8.1.dist-info/METADATA,sha256=LQR8kK-wQSM9bqc_8WM0SUmYXYGrsaoHB5_GBFeECSg,13192
-histoseg-0.1.8.1.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-histoseg-0.1.8.1.dist-info/top_level.txt,sha256=dvphnIeqaZamvJULm-I5qztYoGO8WLLCY85z0xrXsb0,9
-histoseg-0.1.8.1.dist-info/RECORD,,
+histoseg-0.1.9.dist-info/licenses/LICENSE,sha256=z7Ztufk460DPfU3rgZEstjCQK3EbwHbD4JSmF_7y0qA,4764
+histoseg-0.1.9.dist-info/METADATA,sha256=TunvQghh-VpDRKWo8ZXgXslZLLHh5rsZRd2mCxQym7o,13239
+histoseg-0.1.9.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+histoseg-0.1.9.dist-info/top_level.txt,sha256=dvphnIeqaZamvJULm-I5qztYoGO8WLLCY85z0xrXsb0,9
+histoseg-0.1.9.dist-info/RECORD,,