marsilea 0.4.2__py3-none-any.whl → 0.4.4__py3-none-any.whl

marsilea/__init__.py

@@ -1,6 +1,6 @@
 """Declarative creation of composable visualization"""
 
-__version__ = "0.4.2"
+__version__ = "0.4.4"
 
 import marsilea.plotter as plotter
 from ._deform import Deformation

marsilea/_deform.py

@@ -84,19 +84,29 @@ class Deformation:
         self.data_col_reindex = reindex
         self._col_clustered = False
 
-    def set_cluster(self, col=None, row=None, use_meta=True, linkage=None, **kwargs):
+    def set_cluster(
+        self,
+        col=None,
+        row=None,
+        use_meta=True,
+        linkage=None,
+        meta_linkage=None,
+        **kwargs,
+    ):
         if col is not None:
             self.is_col_cluster = col
             self.col_cluster_kws = kwargs
             self._col_clustered = False
             self._use_col_meta = use_meta
             self.col_linkage = linkage
+            self.col_meta_linkage = meta_linkage
         if row is not None:
             self.is_row_cluster = row
             self.row_cluster_kws = kwargs
             self._row_clustered = False
             self._use_row_meta = use_meta
             self.row_linkage = linkage
+            self.row_meta_linkage = meta_linkage
 
     def get_data(self):
         data = self.data
@@ -204,7 +214,9 @@ class Deformation:
                     Dendrogram(chunk, linkage=linkage, key=k, **self.row_cluster_kws)
                 )
 
-            dg = GroupDendrogram(dens, **self.row_cluster_kws)
+            dg = GroupDendrogram(
+                dens, linkage=self.row_meta_linkage, **self.row_cluster_kws
+            )
             if self._use_row_meta:
                 self.row_chunk_index = dg.reorder_index
             else:
@@ -232,7 +244,9 @@ class Deformation:
                 dens.append(
                     Dendrogram(chunk.T, linkage=linkage, key=k, **self.col_cluster_kws)
                 )
-            dg = GroupDendrogram(dens, **self.col_cluster_kws)
+            dg = GroupDendrogram(
+                dens, linkage=self.col_meta_linkage, **self.col_cluster_kws
+            )
             if self._use_col_meta:
                 self.col_chunk_index = dg.reorder_index
             else:

marsilea/base.py

@@ -606,9 +606,15 @@ class WhiteBoard(LegendMaker):
         return compose_board
 
     def _freeze_flex_plots(self, figure):
+        main_cell = self.layout.main_cell
+        main_width = main_cell.width
+        main_height = main_cell.height
+
         for plan in self._col_plan + self._row_plan:
             if plan.size is None:
-                render_size = plan.get_canvas_size(figure)
+                render_size = plan.get_canvas_size(
+                    figure, main_width=main_width, main_height=main_height
+                )
                 if render_size is not None:
                     self.layout.set_render_size(plan.name, render_size)
 
@@ -835,6 +841,7 @@ class ClusterBoard(WhiteBoard):
         method=None,
         metric=None,
         linkage=None,
+        meta_linkage=None,
         add_meta=True,
         add_base=True,
         add_divider=True,
@@ -848,6 +855,7 @@ class ClusterBoard(WhiteBoard):
         size=0.5,
         pad=0.0,
         get_meta_center=None,
+        rasterized=False,
     ):
         """Run cluster and add dendrogram
 
@@ -869,6 +877,10 @@ class ClusterBoard(WhiteBoard):
             Precomputed linkage matrix.
             See scipy's :meth:`linkage <scipy.cluster.hierarchy.linkage>` for
             specific format.
+        meta_linkage : ndarray
+            Precomputed chunk-level linkage matrix.
+            See scipy's :meth:`linkage <scipy.cluster.hierarchy.linkage>` for
+            specific format.
         add_meta : None | bool
             By default, add_meta is set to False if the linkage is provided, otherwise True.
             If the data is split, a meta dendrogram can be drawn for data
@@ -905,6 +917,8 @@ class ClusterBoard(WhiteBoard):
             array as input and return a 1D numpy array of the same length as the number
             of columns in the input, representing the centroid. The default will use the
             mean values.
+        rasterized : bool
+            If True, the dendrogram will be rasterized
 
         Examples
         --------
@@ -926,7 +940,7 @@ class ClusterBoard(WhiteBoard):
             :context: close-figs
 
             >>> h = ma.Heatmap(data)
-            >>> h.hsplit(cut=[4, 8])
+            >>> h.cut_rows(cut=[4, 8])
             >>> h.add_dendrogram("left", add_base=False)
             >>> h.render()
 
@@ -936,7 +950,7 @@ class ClusterBoard(WhiteBoard):
             :context: close-figs
 
             >>> h = ma.Heatmap(data)
-            >>> h.hsplit(cut=[4, 8])
+            >>> h.cut_rows(cut=[4, 8])
             >>> h.add_dendrogram("left", colors=["#5470c6", "#91cc75", "#fac858"])
             >>> h.render()
 
@@ -974,6 +988,7 @@ class ClusterBoard(WhiteBoard):
             colors=colors,
             divider_style=divider_style,
             meta_ratio=meta_ratio,
+            rasterized=rasterized,
         )
 
         deform = self.get_deform()
@@ -985,8 +1000,10 @@ class ClusterBoard(WhiteBoard):
                 method=method,
                 metric=metric,
                 linkage=linkage,
+                meta_linkage=meta_linkage,
                 use_meta=add_meta,
                 get_meta_center=get_meta_center,
+                rasterized=rasterized,
             )
         else:
             den_options["pos"] = "col"
@@ -996,8 +1013,10 @@ class ClusterBoard(WhiteBoard):
                 method=method,
                 metric=metric,
                 linkage=linkage,
+                meta_linkage=meta_linkage,
                 use_meta=add_meta,
                 get_meta_center=get_meta_center,
+                rasterized=rasterized,
             )
 
     def hsplit(self, cut=None, labels=None, order=None, spacing=0.01):
@@ -1160,7 +1179,7 @@ class ClusterBoard(WhiteBoard):
             >>> h = ma.Heatmap(data)
             >>> labels = ["A", "B", "C", "A", "B", "C", "A", "B", "C", "A"]
             >>> h.group_rows(labels, order=["A", "B", "C"])
-            >>> h.add_left(ma.plotter.Labels(labels), pad=.1)
+            >>> h.add_left(ma.plotter.Labels(labels), pad=0.1)
             >>> h.render()
 
         """
@@ -1202,7 +1221,7 @@ class ClusterBoard(WhiteBoard):
             >>> h = ma.Heatmap(data)
             >>> labels = ["A", "B", "C", "A", "B", "C", "A", "B", "C", "A"]
             >>> h.group_cols(labels, order=["A", "B", "C"])
-            >>> h.add_top(ma.plotter.Labels(labels), pad=.1)
+            >>> h.add_top(ma.plotter.Labels(labels), pad=0.1)
             >>> h.render()
 
         """
@@ -1334,7 +1353,11 @@ class ClusterBoard(WhiteBoard):
                     if (color is not None) & (not is_color_like(color)):
                         color = color[0]
                     den_obj.draw(
-                        ax, orient=den["side"], color=color, linewidth=den["linewidth"]
+                        ax,
+                        orient=den["side"],
+                        color=color,
+                        linewidth=den["linewidth"],
+                        rasterized=den["rasterized"],
                     )
                 else:
                     den_obj.draw(
@@ -1349,6 +1372,7 @@ class ClusterBoard(WhiteBoard):
                         divide=den["add_divider"],
                         divide_style=den["divider_style"],
                         meta_ratio=den["meta_ratio"],
+                        rasterized=den["rasterized"],
                     )
 
     def _render_plan(self):

marsilea/dendrogram.py

@@ -18,6 +18,7 @@ class _DendrogramBase:
         linkage=None,
         get_meta_center=None,
         key=None,
+        **kwargs,
     ):
         self.key = key
         self.data = data
@@ -137,7 +138,9 @@ class _DendrogramBase:
         y1 = yc[1]
         return x1, y1
 
-    def _draw_dendrogram(self, ax, orient="top", color=".1", linewidth=0.7):
+    def _draw_dendrogram(
+        self, ax, orient="top", color=".1", linewidth=0.7, rasterized=False
+    ):
         x_coords = self._render_x_coords
         y_coords = self._render_y_coords
         if orient in ["right", "left"]:
@@ -147,6 +150,7 @@ class _DendrogramBase:
             [list(zip(x, y)) for x, y in zip(x_coords, y_coords)],
             color=color,
             linewidth=linewidth,
+            rasterized=rasterized,
         )
         ax.add_collection(lines)
 
@@ -173,6 +177,7 @@ class Dendrogram(_DendrogramBase):
         linkage=None,
         get_meta_center=None,
         key=None,
+        **kwargs,
     ):
         super().__init__(
             data,
@@ -181,6 +186,7 @@ class Dendrogram(_DendrogramBase):
             key=key,
             linkage=linkage,
             get_meta_center=get_meta_center,
+            **kwargs,
         )
 
     # here we left an empty **kwargs to align api with GroupDendrogram
@@ -193,6 +199,7 @@ class Dendrogram(_DendrogramBase):
         add_root=False,
         root_color=None,
         control_ax=True,
+        rasterized=False,
         **kwargs,
     ):
         """
@@ -210,6 +217,8 @@ class Dendrogram(_DendrogramBase):
             The color of the root line
         control_ax : bool
             Adjust the axes to ensure the dendrogram will display correctly
+        rasterized : bool
+            Rasterize the dendrogram to speed up rendering
 
         Returns
         -------
@@ -219,7 +228,9 @@ class Dendrogram(_DendrogramBase):
         root_color = color if root_color is None else root_color
         linewidth = 0.7 if linewidth is None else linewidth
 
-        self._draw_dendrogram(ax, orient=orient, color=color, linewidth=linewidth)
+        self._draw_dendrogram(
+            ax, orient=orient, color=color, linewidth=linewidth, rasterized=rasterized
+        )
 
         xlim = self._render_xlim
         ylim = self._render_ylim
@@ -244,7 +255,11 @@ class Dendrogram(_DendrogramBase):
                 x2 = x1
                 y2 = ylim[1]
             root_line = Line2D(
-                [x1, x2], [y1, y2], color=root_color, linewidth=linewidth
+                [x1, x2],
+                [y1, y2],
+                color=root_color,
+                linewidth=linewidth,
+                rasterized=rasterized,
             )
             ax.add_artist(root_line)
 
@@ -267,13 +282,20 @@ class GroupDendrogram(_DendrogramBase):
         dens: List[Dendrogram],
         method=None,
         metric=None,
+        linkage=None,
         get_meta_center=None,
         key=None,
         **kwargs,
     ):
         data = np.vstack([d.center for d in dens])
         super().__init__(
-            data, method=method, metric=metric, get_meta_center=get_meta_center, key=key
+            data,
+            method=method,
+            metric=metric,
+            linkage=linkage,
+            get_meta_center=get_meta_center,
+            key=key,
+            **kwargs,
         )
         self.orig_dens = np.asarray(dens)
         self.dens = np.asarray(dens)[self.reorder_index]
@@ -305,6 +327,7 @@ class GroupDendrogram(_DendrogramBase):
         divide=True,
         divide_style="--",
         meta_ratio=0.2,
+        rasterized=False,
     ):
         """
 
@@ -330,6 +353,8 @@ class GroupDendrogram(_DendrogramBase):
             The linestyle of the divide line
         meta_ratio : float
             The size of meta dendrogram relative to the base dendrogram
+        rasterized : bool
+            Rasterize the dendrogram to speed up rendering
 
         """
 
@@ -418,7 +443,11 @@ class GroupDendrogram(_DendrogramBase):
         if add_meta:
             # Add meta dendrogram
             self._draw_dendrogram(
-                ax, orient=orient, color=meta_color, linewidth=linewidth
+                ax,
+                orient=orient,
+                color=meta_color,
+                linewidth=linewidth,
+                rasterized=rasterized,
             )
 
         if divide & add_base & add_meta:
@@ -432,6 +461,7 @@ class GroupDendrogram(_DendrogramBase):
                     linestyles=divide_style,
                     color=meta_color,
                     linewidth=linewidth,
+                    rasterized=rasterized,
                 )
             else:
                 ax.vlines(
@@ -441,6 +471,7 @@ class GroupDendrogram(_DendrogramBase):
                     linestyles=divide_style,
                     color=meta_color,
                     linewidth=linewidth,
+                    rasterized=rasterized,
                 )
 
         if add_base:
@@ -455,6 +486,7 @@ class GroupDendrogram(_DendrogramBase):
                         linewidth=linewidth,
                         root_color=meta_color,
                         control_ax=False,
+                        rasterized=rasterized,
                     )
 
         xlim = render_xlim

marsilea/exceptions.py

@@ -7,15 +7,16 @@ class DuplicateName(Exception):
 
 
 class DuplicatePlotter(Exception):
-
     def __init__(self, plotter):
         self.plotter = plotter
 
     def __str__(self):
         name = self.plotter.__class__.__name__
-        return (f"You have added `{name}` "
-                f"to the `{self.plotter.side}`, "
-                f"please create a new `{name}` if you want to add it again.")
+        return (
+            f"The `{name}` added"
+            f"to the `{self.plotter.side}` has been registered,"
+            f"please create a new `{name}` if you want to add the same plotter again."
+        )
 
 
 class SplitTwice(Exception):

marsilea/heatmap.py

@@ -50,6 +50,7 @@ class Heatmap(ClusterBoard):
         height=None,
         cluster_data=None,
         init_main=True,
+        **kwargs,
     ):
         if cluster_data is None:
             if isinstance(data, pd.DataFrame):
@@ -75,6 +76,7 @@ class Heatmap(ClusterBoard):
             annot_kws=annot_kws,
             label=label,
             cbar_kws=cbar_kws,
+            **kwargs,
         )
         name = get_plot_name(name, "main", mesh.__class__.__name__)
         mesh.set(name=name)

marsilea/plotter/__init__.py

@@ -23,6 +23,8 @@ __all__ = [
     "Chunk",
     "FixedChunk",
     "Area",
+    "Image",
+    "Emoji",
 ]
 
 from ._seaborn import Bar, Box, Boxen, Violin, Point, Strip, Swarm
@@ -33,5 +35,5 @@ from .bio import SeqLogo
 from .mesh import Colors, ColorMesh, SizedMesh, MarkerMesh, TextMesh
 from .text import Labels, AnnoLabels, Title, Chunk, FixedChunk
 
-# from ._images import Emoji
+from .images import Emoji, Image
 from .area import Area

marsilea/plotter/_seaborn.py

@@ -96,7 +96,7 @@ class _SeabornBase(StatsBase):
                 df["hue"] = hue
                 dfs.append(df)
 
-            pdata = pd.concat(dfs)
+            pdata = pd.concat(dfs).reset_index(drop=True)
             self.kws["hue"] = "hue"
             self.kws["hue_order"] = self.hue
             if self.get_orient() == "h":
@@ -123,7 +123,8 @@ class _SeabornBase(StatsBase):
             ax.invert_xaxis()
         # barplot(data=data, orient=orient, ax=ax, **self.kws)
         plotter = getattr(seaborn, self._seaborn_plot)
-
+        print(pdata)
+        print(options)
         plotter(data=pdata, orient=orient, ax=ax, **options)
         ax.set(xlabel=None, ylabel=None)
         leg = ax.get_legend()
@@ -195,7 +196,7 @@ def _seaborn_doc(obj: _SeabornBase):
         >>> sdata = {sdata}
         >>> plot = {cls_name}(sdata, {kws})
         >>> h = ma.Heatmap(data)
-        >>> h.hsplit(cut=[3, 7])
+        >>> h.cut_rows(cut=[3, 7])
         >>> h.add_right(plot)
         >>> h.render()
     """
@@ -208,7 +209,7 @@ def _seaborn_doc(obj: _SeabornBase):
         
         >>> plot = {cls_name}({hue_data}, {kws})
         >>> h = ma.Heatmap(data)
-        >>> h.hsplit(cut=[3, 7])
+        >>> h.cut_rows(cut=[3, 7])
         >>> h.add_right(plot)
         >>> h.render()
         
@@ -222,7 +223,7 @@ def _seaborn_doc(obj: _SeabornBase):
         >>> anno = ma.plotter.Chunk(['C1', 'C2', 'C3'], colors, padding=10)
         >>> cb = ma.ClusterBoard(data, height=2, margin=.5)
         >>> cb.add_layer(plot)
-        >>> cb.vsplit(cut=[3, 7])
+        >>> cb.cut_cols([3, 7])
         >>> cb.add_bottom(anno)
         >>> cb.render()
         
@@ -233,9 +234,10 @@ def _seaborn_doc(obj: _SeabornBase):
         
         >>> plot = {cls_name}(sdata, orient='h',
         ...                   {h_kws})
+        >>> anno = ma.plotter.Chunk(['C1', 'C2', 'C3'], colors, padding=10)
         >>> cb = ma.ClusterBoard(data.T, width=2)
         >>> cb.add_layer(plot)
-        >>> cb.hsplit(cut=[3, 7])
+        >>> cb.cut_rows([3, 7])
         >>> cb.add_left(anno)
         >>> cb.render()
         

marsilea/plotter/arc.py

@@ -141,8 +141,7 @@ class Arc(StatsBase):
         >>> colors = ["C0", "C1", "C2", "C3", "C4", "C5", "C6"]
         >>> labels = ["A", "B", "C", "D", "E", "F", "G"]
         >>> h = ma.Heatmap(np.random.rand(10, 10))
-        >>> h.add_top(Arc(anchors, links, weights=weights,
-        ...               colors=colors, labels=labels))
+        >>> h.add_top(Arc(anchors, links, weights=weights, colors=colors, labels=labels))
         >>> h.render()
 
 

marsilea/plotter/bar.py

@@ -276,13 +276,13 @@ class CenterBar(_BarBase):
 class StackBar(_BarBase):
     """Stacked Bar
 
-     Parameters
-     ----------
-     data : np.ndarray, pd.DataFrame
+    Parameters
+    ----------
+    data : np.ndarray, pd.DataFrame
         2D data, index of dataframe is used as the name of items.
-     items : list of str
+    items : list of str
         The name of items.
-     colors : list of colors, mapping of (item, color)
+    colors : list of colors, mapping of (item, color)
         The colors of the bar for each item.
     orient : {"v", "h"}
         The orientation of the plot
@@ -309,8 +309,9 @@ class StackBar(_BarBase):
         :context: close-figs
 
         >>> from marsilea.plotter import StackBar
-        >>> stack_data = pd.DataFrame(data=np.random.randint(1, 10, (5, 10)),
-        ...                           index=list("abcde"))
+        >>> stack_data = pd.DataFrame(
+        ...     data=np.random.randint(1, 10, (5, 10)), index=list("abcde")
+        ... )
         >>> _, ax = plt.subplots()
         >>> StackBar(stack_data).render(ax)
 

marsilea/plotter/base.py

@@ -14,6 +14,13 @@ from .._deform import Deformation
 from ..exceptions import DataError, SplitConflict
 
 
+# class DataValidator:
+#
+#     @singledispatch(np.ndarray)
+#     def parse(self, data):
+#         pass
+
+
 class DataLoader:
     """Handle user data"""
 
@@ -397,7 +404,8 @@ class RenderPlan:
                 return self._get_intact_render_spec(axes)
         except Exception as _:
             raise DataError(
-                f"Please check your data input with {self.__class__.__name__}"
+                f"Please check your data input "
+                f"with {self.__class__.__name__} at '{self.side}'"
             )
 
     # def get_render_data(self):
@@ -508,7 +516,9 @@ class RenderPlan:
             if self._plan_label is not None:
                 self._plan_label.add(axes, self.side)
 
-    def get_canvas_size(self, figure) -> float:
+    def get_canvas_size(
+        self, figure, main_height=None, main_width=None, **kwargs
+    ) -> float:
         """
         If the size is unknown before rendering, this function must be
         implemented to return the canvas size in inches.

marsilea/plotter/bio.py

@@ -102,8 +102,9 @@ class SeqLogo(StatsBase):
 
         >>> import pandas as pd
         >>> from marsilea.plotter import SeqLogo
-        >>> matrix = pd.DataFrame(data=np.random.randint(1, 10, (4, 10)),
-        ...                       index=list("ACGT"))
+        >>> matrix = pd.DataFrame(
+        ...     data=np.random.randint(1, 10, (4, 10)), index=list("ACGT")
+        ... )
         >>> _, ax = plt.subplots()
         >>> colors = {"A": "r", "C": "b", "G": "g", "T": "black"}
         >>> SeqLogo(matrix, color_encode=colors).render(ax)