"""Functions to visualize matrices of data.""" import warnings import matplotlib as mpl from matplotlib.collections import LineCollection import matplotlib.pyplot as plt from matplotlib import gridspec import numpy as np import pandas as pd try: from scipy.cluster import hierarchy _no_scipy = False except ImportError: _no_scipy = True from . import cm from .axisgrid import Grid from ._compat import get_colormap from .utils import ( despine, axis_ticklabels_overlap, relative_luminance, to_utf8, _draw_figure, ) __all__ = ["heatmap", "clustermap"] def _index_to_label(index): """Convert a pandas index or multiindex to an axis label.""" if isinstance(index, pd.MultiIndex): return "-".join(map(to_utf8, index.names)) else: return index.name def _index_to_ticklabels(index): """Convert a pandas index or multiindex into ticklabels.""" if isinstance(index, pd.MultiIndex): return ["-".join(map(to_utf8, i)) for i in index.values] else: return index.values def _convert_colors(colors): """Convert either a list of colors or nested lists of colors to RGB.""" to_rgb = mpl.colors.to_rgb try: to_rgb(colors[0]) # If this works, there is only one level of colors return list(map(to_rgb, colors)) except ValueError: # If we get here, we have nested lists return [list(map(to_rgb, color_list)) for color_list in colors] def _matrix_mask(data, mask): """Ensure that data and mask are compatible and add missing values. Values will be plotted for cells where ``mask`` is ``False``. ``data`` is expected to be a DataFrame; ``mask`` can be an array or a DataFrame. """ if mask is None: mask = np.zeros(data.shape, bool) if isinstance(mask, np.ndarray): # For array masks, ensure that shape matches data then convert if mask.shape != data.shape: raise ValueError("Mask must have the same shape as data.") mask = pd.DataFrame(mask, index=data.index, columns=data.columns, dtype=bool) elif isinstance(mask, pd.DataFrame): # For DataFrame masks, ensure that semantic labels match data if not mask.index.equals(data.index) \ and mask.columns.equals(data.columns): err = "Mask must have the same index and columns as data." raise ValueError(err) # Add any cells with missing data to the mask # This works around an issue where `plt.pcolormesh` doesn't represent # missing data properly mask = mask | pd.isnull(data) return mask class _HeatMapper: """Draw a heatmap plot of a matrix with nice labels and colormaps.""" def __init__(self, data, vmin, vmax, cmap, center, robust, annot, fmt, annot_kws, cbar, cbar_kws, xticklabels=True, yticklabels=True, mask=None): """Initialize the plotting object.""" # We always want to have a DataFrame with semantic information # and an ndarray to pass to matplotlib if isinstance(data, pd.DataFrame): plot_data = data.values else: plot_data = np.asarray(data) data = pd.DataFrame(plot_data) # Validate the mask and convert to DataFrame mask = _matrix_mask(data, mask) plot_data = np.ma.masked_where(np.asarray(mask), plot_data) # Get good names for the rows and columns xtickevery = 1 if isinstance(xticklabels, int): xtickevery = xticklabels xticklabels = _index_to_ticklabels(data.columns) elif xticklabels is True: xticklabels = _index_to_ticklabels(data.columns) elif xticklabels is False: xticklabels = [] ytickevery = 1 if isinstance(yticklabels, int): ytickevery = yticklabels yticklabels = _index_to_ticklabels(data.index) elif yticklabels is True: yticklabels = _index_to_ticklabels(data.index) elif yticklabels is False: yticklabels = [] if not len(xticklabels): self.xticks = [] self.xticklabels = [] elif isinstance(xticklabels, str) and xticklabels == "auto": self.xticks = "auto" self.xticklabels = _index_to_ticklabels(data.columns) else: self.xticks, self.xticklabels = self._skip_ticks(xticklabels, xtickevery) if not len(yticklabels): self.yticks = [] self.yticklabels = [] elif isinstance(yticklabels, str) and yticklabels == "auto": self.yticks = "auto" self.yticklabels = _index_to_ticklabels(data.index) else: self.yticks, self.yticklabels = self._skip_ticks(yticklabels, ytickevery) # Get good names for the axis labels xlabel = _index_to_label(data.columns) ylabel = _index_to_label(data.index) self.xlabel = xlabel if xlabel is not None else "" self.ylabel = ylabel if ylabel is not None else "" # Determine good default values for the colormapping self._determine_cmap_params(plot_data, vmin, vmax, cmap, center, robust) # Sort out the annotations if annot is None or annot is False: annot = False annot_data = None else: if isinstance(annot, bool): annot_data = plot_data else: annot_data = np.asarray(annot) if annot_data.shape != plot_data.shape: err = "`data` and `annot` must have same shape." raise ValueError(err) annot = True # Save other attributes to the object self.data = data self.plot_data = plot_data self.annot = annot self.annot_data = annot_data self.fmt = fmt self.annot_kws = {} if annot_kws is None else annot_kws.copy() self.cbar = cbar self.cbar_kws = {} if cbar_kws is None else cbar_kws.copy() def _determine_cmap_params(self, plot_data, vmin, vmax, cmap, center, robust): """Use some heuristics to set good defaults for colorbar and range.""" # plot_data is a np.ma.array instance calc_data = plot_data.astype(float).filled(np.nan) if vmin is None: if robust: vmin = np.nanpercentile(calc_data, 2) else: vmin = np.nanmin(calc_data) if vmax is None: if robust: vmax = np.nanpercentile(calc_data, 98) else: vmax = np.nanmax(calc_data) self.vmin, self.vmax = vmin, vmax # Choose default colormaps if not provided if cmap is None: if center is None: self.cmap = cm.rocket else: self.cmap = cm.icefire elif isinstance(cmap, str): self.cmap = get_colormap(cmap) elif isinstance(cmap, list): self.cmap = mpl.colors.ListedColormap(cmap) else: self.cmap = cmap # Recenter a divergent colormap if center is not None: # Copy bad values # in mpl<3.2 only masked values are honored with "bad" color spec # (see https://github.com/matplotlib/matplotlib/pull/14257) bad = self.cmap(np.ma.masked_invalid([np.nan]))[0] # under/over values are set for sure when cmap extremes # do not map to the same color as +-inf under = self.cmap(-np.inf) over = self.cmap(np.inf) under_set = under != self.cmap(0) over_set = over != self.cmap(self.cmap.N - 1) vrange = max(vmax - center, center - vmin) normlize = mpl.colors.Normalize(center - vrange, center + vrange) cmin, cmax = normlize([vmin, vmax]) cc = np.linspace(cmin, cmax, 256) self.cmap = mpl.colors.ListedColormap(self.cmap(cc)) self.cmap.set_bad(bad) if under_set: self.cmap.set_under(under) if over_set: self.cmap.set_over(over) def _annotate_heatmap(self, ax, mesh): """Add textual labels with the value in each cell.""" mesh.update_scalarmappable() height, width = self.annot_data.shape xpos, ypos = np.meshgrid(np.arange(width) + .5, np.arange(height) + .5) for x, y, m, color, val in zip(xpos.flat, ypos.flat, mesh.get_array().flat, mesh.get_facecolors(), self.annot_data.flat): if m is not np.ma.masked: lum = relative_luminance(color) text_color = ".15" if lum > .408 else "w" annotation = ("{:" + self.fmt + "}").format(val) text_kwargs = dict(color=text_color, ha="center", va="center") text_kwargs.update(self.annot_kws) ax.text(x, y, annotation, **text_kwargs) def _skip_ticks(self, labels, tickevery): """Return ticks and labels at evenly spaced intervals.""" n = len(labels) if tickevery == 0: ticks, labels = [], [] elif tickevery == 1: ticks, labels = np.arange(n) + .5, labels else: start, end, step = 0, n, tickevery ticks = np.arange(start, end, step) + .5 labels = labels[start:end:step] return ticks, labels def _auto_ticks(self, ax, labels, axis): """Determine ticks and ticklabels that minimize overlap.""" transform = ax.figure.dpi_scale_trans.inverted() bbox = ax.get_window_extent().transformed(transform) size = [bbox.width, bbox.height][axis] axis = [ax.xaxis, ax.yaxis][axis] tick, = axis.set_ticks([0]) fontsize = tick.label1.get_size() max_ticks = int(size // (fontsize / 72)) if max_ticks < 1: return [], [] tick_every = len(labels) // max_ticks + 1 tick_every = 1 if tick_every == 0 else tick_every ticks, labels = self._skip_ticks(labels, tick_every) return ticks, labels def plot(self, ax, cax, kws): """Draw the heatmap on the provided Axes.""" # Remove all the Axes spines despine(ax=ax, left=True, bottom=True) # setting vmin/vmax in addition to norm is deprecated # so avoid setting if norm is set if kws.get("norm") is None: kws.setdefault("vmin", self.vmin) kws.setdefault("vmax", self.vmax) # Draw the heatmap mesh = ax.pcolormesh(self.plot_data, cmap=self.cmap, **kws) # Set the axis limits ax.set(xlim=(0, self.data.shape[1]), ylim=(0, self.data.shape[0])) # Invert the y axis to show the plot in matrix form ax.invert_yaxis() # Possibly add a colorbar if self.cbar: cb = ax.figure.colorbar(mesh, cax, ax, **self.cbar_kws) cb.outline.set_linewidth(0) # If rasterized is passed to pcolormesh, also rasterize the # colorbar to avoid white lines on the PDF rendering if kws.get('rasterized', False): cb.solids.set_rasterized(True) # Add row and column labels if isinstance(self.xticks, str) and self.xticks == "auto": xticks, xticklabels = self._auto_ticks(ax, self.xticklabels, 0) else: xticks, xticklabels = self.xticks, self.xticklabels if isinstance(self.yticks, str) and self.yticks == "auto": yticks, yticklabels = self._auto_ticks(ax, self.yticklabels, 1) else: yticks, yticklabels = self.yticks, self.yticklabels ax.set(xticks=xticks, yticks=yticks) xtl = ax.set_xticklabels(xticklabels) ytl = ax.set_yticklabels(yticklabels, rotation="vertical") plt.setp(ytl, va="center") # GH2484 # Possibly rotate them if they overlap _draw_figure(ax.figure) if axis_ticklabels_overlap(xtl): plt.setp(xtl, rotation="vertical") if axis_ticklabels_overlap(ytl): plt.setp(ytl, rotation="horizontal") # Add the axis labels ax.set(xlabel=self.xlabel, ylabel=self.ylabel) # Annotate the cells with the formatted values if self.annot: self._annotate_heatmap(ax, mesh) def heatmap( data, *, vmin=None, vmax=None, cmap=None, center=None, robust=False, annot=None, fmt=".2g", annot_kws=None, linewidths=0, linecolor="white", cbar=True, cbar_kws=None, cbar_ax=None, square=False, xticklabels="auto", yticklabels="auto", mask=None, ax=None, **kwargs ): """Plot rectangular data as a color-encoded matrix. This is an Axes-level function and will draw the heatmap into the currently-active Axes if none is provided to the ``ax`` argument. Part of this Axes space will be taken and used to plot a colormap, unless ``cbar`` is False or a separate Axes is provided to ``cbar_ax``. Parameters ---------- data : rectangular dataset 2D dataset that can be coerced into an ndarray. If a Pandas DataFrame is provided, the index/column information will be used to label the columns and rows. vmin, vmax : floats, optional Values to anchor the colormap, otherwise they are inferred from the data and other keyword arguments. cmap : matplotlib colormap name or object, or list of colors, optional The mapping from data values to color space. If not provided, the default will depend on whether ``center`` is set. center : float, optional The value at which to center the colormap when plotting divergent data. Using this parameter will change the default ``cmap`` if none is specified. robust : bool, optional If True and ``vmin`` or ``vmax`` are absent, the colormap range is computed with robust quantiles instead of the extreme values. annot : bool or rectangular dataset, optional If True, write the data value in each cell. If an array-like with the same shape as ``data``, then use this to annotate the heatmap instead of the data. Note that DataFrames will match on position, not index. fmt : str, optional String formatting code to use when adding annotations. annot_kws : dict of key, value mappings, optional Keyword arguments for :meth:`matplotlib.axes.Axes.text` when ``annot`` is True. linewidths : float, optional Width of the lines that will divide each cell. linecolor : color, optional Color of the lines that will divide each cell. cbar : bool, optional Whether to draw a colorbar. cbar_kws : dict of key, value mappings, optional Keyword arguments for :meth:`matplotlib.figure.Figure.colorbar`. cbar_ax : matplotlib Axes, optional Axes in which to draw the colorbar, otherwise take space from the main Axes. square : bool, optional If True, set the Axes aspect to "equal" so each cell will be square-shaped. xticklabels, yticklabels : "auto", bool, list-like, or int, optional If True, plot the column names of the dataframe. If False, don't plot the column names. If list-like, plot these alternate labels as the xticklabels. If an integer, use the column names but plot only every n label. If "auto", try to densely plot non-overlapping labels. mask : bool array or DataFrame, optional If passed, data will not be shown in cells where ``mask`` is True. Cells with missing values are automatically masked. ax : matplotlib Axes, optional Axes in which to draw the plot, otherwise use the currently-active Axes. kwargs : other keyword arguments All other keyword arguments are passed to :meth:`matplotlib.axes.Axes.pcolormesh`. Returns ------- ax : matplotlib Axes Axes object with the heatmap. See Also -------- clustermap : Plot a matrix using hierarchical clustering to arrange the rows and columns. Examples -------- .. include:: ../docstrings/heatmap.rst """ # Initialize the plotter object plotter = _HeatMapper(data, vmin, vmax, cmap, center, robust, annot, fmt, annot_kws, cbar, cbar_kws, xticklabels, yticklabels, mask) # Add the pcolormesh kwargs here kwargs["linewidths"] = linewidths kwargs["edgecolor"] = linecolor # Draw the plot and return the Axes if ax is None: ax = plt.gca() if square: ax.set_aspect("equal") plotter.plot(ax, cbar_ax, kwargs) return ax class _DendrogramPlotter: """Object for drawing tree of similarities between data rows/columns""" def __init__(self, data, linkage, metric, method, axis, label, rotate): """Plot a dendrogram of the relationships between the columns of data Parameters ---------- data : pandas.DataFrame Rectangular data """ self.axis = axis if self.axis == 1: data = data.T if isinstance(data, pd.DataFrame): array = data.values else: array = np.asarray(data) data = pd.DataFrame(array) self.array = array self.data = data self.shape = self.data.shape self.metric = metric self.method = method self.axis = axis self.label = label self.rotate = rotate if linkage is None: self.linkage = self.calculated_linkage else: self.linkage = linkage self.dendrogram = self.calculate_dendrogram() # Dendrogram ends are always at multiples of 5, who knows why ticks = 10 * np.arange(self.data.shape[0]) + 5 if self.label: ticklabels = _index_to_ticklabels(self.data.index) ticklabels = [ticklabels[i] for i in self.reordered_ind] if self.rotate: self.xticks = [] self.yticks = ticks self.xticklabels = [] self.yticklabels = ticklabels self.ylabel = _index_to_label(self.data.index) self.xlabel = '' else: self.xticks = ticks self.yticks = [] self.xticklabels = ticklabels self.yticklabels = [] self.ylabel = '' self.xlabel = _index_to_label(self.data.index) else: self.xticks, self.yticks = [], [] self.yticklabels, self.xticklabels = [], [] self.xlabel, self.ylabel = '', '' self.dependent_coord = self.dendrogram['dcoord'] self.independent_coord = self.dendrogram['icoord'] def _calculate_linkage_scipy(self): linkage = hierarchy.linkage(self.array, method=self.method, metric=self.metric) return linkage def _calculate_linkage_fastcluster(self): import fastcluster # Fastcluster has a memory-saving vectorized version, but only # with certain linkage methods, and mostly with euclidean metric # vector_methods = ('single', 'centroid', 'median', 'ward') euclidean_methods = ('centroid', 'median', 'ward') euclidean = self.metric == 'euclidean' and self.method in \ euclidean_methods if euclidean or self.method == 'single': return fastcluster.linkage_vector(self.array, method=self.method, metric=self.metric) else: linkage = fastcluster.linkage(self.array, method=self.method, metric=self.metric) return linkage @property def calculated_linkage(self): try: return self._calculate_linkage_fastcluster() except ImportError: if np.prod(self.shape) >= 10000: msg = ("Clustering large matrix with scipy. Installing " "`fastcluster` may give better performance.") warnings.warn(msg) return self._calculate_linkage_scipy() def calculate_dendrogram(self): """Calculates a dendrogram based on the linkage matrix Made a separate function, not a property because don't want to recalculate the dendrogram every time it is accessed. Returns ------- dendrogram : dict Dendrogram dictionary as returned by scipy.cluster.hierarchy .dendrogram. The important key-value pairing is "reordered_ind" which indicates the re-ordering of the matrix """ return hierarchy.dendrogram(self.linkage, no_plot=True, color_threshold=-np.inf) @property def reordered_ind(self): """Indices of the matrix, reordered by the dendrogram""" return self.dendrogram['leaves'] def plot(self, ax, tree_kws): """Plots a dendrogram of the similarities between data on the axes Parameters ---------- ax : matplotlib.axes.Axes Axes object upon which the dendrogram is plotted """ tree_kws = {} if tree_kws is None else tree_kws.copy() tree_kws.setdefault("linewidths", .5) tree_kws.setdefault("colors", tree_kws.pop("color", (.2, .2, .2))) if self.rotate and self.axis == 0: coords = zip(self.dependent_coord, self.independent_coord) else: coords = zip(self.independent_coord, self.dependent_coord) lines = LineCollection([list(zip(x, y)) for x, y in coords], **tree_kws) ax.add_collection(lines) number_of_leaves = len(self.reordered_ind) max_dependent_coord = max(map(max, self.dependent_coord)) if self.rotate: ax.yaxis.set_ticks_position('right') # Constants 10 and 1.05 come from # `scipy.cluster.hierarchy._plot_dendrogram` ax.set_ylim(0, number_of_leaves * 10) ax.set_xlim(0, max_dependent_coord * 1.05) ax.invert_xaxis() ax.invert_yaxis() else: # Constants 10 and 1.05 come from # `scipy.cluster.hierarchy._plot_dendrogram` ax.set_xlim(0, number_of_leaves * 10) ax.set_ylim(0, max_dependent_coord * 1.05) despine(ax=ax, bottom=True, left=True) ax.set(xticks=self.xticks, yticks=self.yticks, xlabel=self.xlabel, ylabel=self.ylabel) xtl = ax.set_xticklabels(self.xticklabels) ytl = ax.set_yticklabels(self.yticklabels, rotation='vertical') # Force a draw of the plot to avoid matplotlib window error _draw_figure(ax.figure) if len(ytl) > 0 and axis_ticklabels_overlap(ytl): plt.setp(ytl, rotation="horizontal") if len(xtl) > 0 and axis_ticklabels_overlap(xtl): plt.setp(xtl, rotation="vertical") return self def dendrogram( data, *, linkage=None, axis=1, label=True, metric='euclidean', method='average', rotate=False, tree_kws=None, ax=None ): """Draw a tree diagram of relationships within a matrix Parameters ---------- data : pandas.DataFrame Rectangular data linkage : numpy.array, optional Linkage matrix axis : int, optional Which axis to use to calculate linkage. 0 is rows, 1 is columns. label : bool, optional If True, label the dendrogram at leaves with column or row names metric : str, optional Distance metric. Anything valid for scipy.spatial.distance.pdist method : str, optional Linkage method to use. Anything valid for scipy.cluster.hierarchy.linkage rotate : bool, optional When plotting the matrix, whether to rotate it 90 degrees counter-clockwise, so the leaves face right tree_kws : dict, optional Keyword arguments for the ``matplotlib.collections.LineCollection`` that is used for plotting the lines of the dendrogram tree. ax : matplotlib axis, optional Axis to plot on, otherwise uses current axis Returns ------- dendrogramplotter : _DendrogramPlotter A Dendrogram plotter object. Notes ----- Access the reordered dendrogram indices with dendrogramplotter.reordered_ind """ if _no_scipy: raise RuntimeError("dendrogram requires scipy to be installed") plotter = _DendrogramPlotter(data, linkage=linkage, axis=axis, metric=metric, method=method, label=label, rotate=rotate) if ax is None: ax = plt.gca() return plotter.plot(ax=ax, tree_kws=tree_kws) class ClusterGrid(Grid): def __init__(self, data, pivot_kws=None, z_score=None, standard_scale=None, figsize=None, row_colors=None, col_colors=None, mask=None, dendrogram_ratio=None, colors_ratio=None, cbar_pos=None): """Grid object for organizing clustered heatmap input on to axes""" if _no_scipy: raise RuntimeError("ClusterGrid requires scipy to be available") if isinstance(data, pd.DataFrame): self.data = data else: self.data = pd.DataFrame(data) self.data2d = self.format_data(self.data, pivot_kws, z_score, standard_scale) self.mask = _matrix_mask(self.data2d, mask) self._figure = plt.figure(figsize=figsize) self.row_colors, self.row_color_labels = \ self._preprocess_colors(data, row_colors, axis=0) self.col_colors, self.col_color_labels = \ self._preprocess_colors(data, col_colors, axis=1) try: row_dendrogram_ratio, col_dendrogram_ratio = dendrogram_ratio except TypeError: row_dendrogram_ratio = col_dendrogram_ratio = dendrogram_ratio try: row_colors_ratio, col_colors_ratio = colors_ratio except TypeError: row_colors_ratio = col_colors_ratio = colors_ratio width_ratios = self.dim_ratios(self.row_colors, row_dendrogram_ratio, row_colors_ratio) height_ratios = self.dim_ratios(self.col_colors, col_dendrogram_ratio, col_colors_ratio) nrows = 2 if self.col_colors is None else 3 ncols = 2 if self.row_colors is None else 3 self.gs = gridspec.GridSpec(nrows, ncols, width_ratios=width_ratios, height_ratios=height_ratios) self.ax_row_dendrogram = self._figure.add_subplot(self.gs[-1, 0]) self.ax_col_dendrogram = self._figure.add_subplot(self.gs[0, -1]) self.ax_row_dendrogram.set_axis_off() self.ax_col_dendrogram.set_axis_off() self.ax_row_colors = None self.ax_col_colors = None if self.row_colors is not None: self.ax_row_colors = self._figure.add_subplot( self.gs[-1, 1]) if self.col_colors is not None: self.ax_col_colors = self._figure.add_subplot( self.gs[1, -1]) self.ax_heatmap = self._figure.add_subplot(self.gs[-1, -1]) if cbar_pos is None: self.ax_cbar = self.cax = None else: # Initialize the colorbar axes in the gridspec so that tight_layout # works. We will move it where it belongs later. This is a hack. self.ax_cbar = self._figure.add_subplot(self.gs[0, 0]) self.cax = self.ax_cbar # Backwards compatibility self.cbar_pos = cbar_pos self.dendrogram_row = None self.dendrogram_col = None def _preprocess_colors(self, data, colors, axis): """Preprocess {row/col}_colors to extract labels and convert colors.""" labels = None if colors is not None: if isinstance(colors, (pd.DataFrame, pd.Series)): # If data is unindexed, raise if (not hasattr(data, "index") and axis == 0) or ( not hasattr(data, "columns") and axis == 1 ): axis_name = "col" if axis else "row" msg = (f"{axis_name}_colors indices can't be matched with data " f"indices. Provide {axis_name}_colors as a non-indexed " "datatype, e.g. by using `.to_numpy()``") raise TypeError(msg) # Ensure colors match data indices if axis == 0: colors = colors.reindex(data.index) else: colors = colors.reindex(data.columns) # Replace na's with white color # TODO We should set these to transparent instead colors = colors.astype(object).fillna('white') # Extract color values and labels from frame/series if isinstance(colors, pd.DataFrame): labels = list(colors.columns) colors = colors.T.values else: if colors.name is None: labels = [""] else: labels = [colors.name] colors = colors.values colors = _convert_colors(colors) return colors, labels def format_data(self, data, pivot_kws, z_score=None, standard_scale=None): """Extract variables from data or use directly.""" # Either the data is already in 2d matrix format, or need to do a pivot if pivot_kws is not None: data2d = data.pivot(**pivot_kws) else: data2d = data if z_score is not None and standard_scale is not None: raise ValueError( 'Cannot perform both z-scoring and standard-scaling on data') if z_score is not None: data2d = self.z_score(data2d, z_score) if standard_scale is not None: data2d = self.standard_scale(data2d, standard_scale) return data2d @staticmethod def z_score(data2d, axis=1): """Standarize the mean and variance of the data axis Parameters ---------- data2d : pandas.DataFrame Data to normalize axis : int Which axis to normalize across. If 0, normalize across rows, if 1, normalize across columns. Returns ------- normalized : pandas.DataFrame Noramlized data with a mean of 0 and variance of 1 across the specified axis. """ if axis == 1: z_scored = data2d else: z_scored = data2d.T z_scored = (z_scored - z_scored.mean()) / z_scored.std() if axis == 1: return z_scored else: return z_scored.T @staticmethod def standard_scale(data2d, axis=1): """Divide the data by the difference between the max and min Parameters ---------- data2d : pandas.DataFrame Data to normalize axis : int Which axis to normalize across. If 0, normalize across rows, if 1, normalize across columns. Returns ------- standardized : pandas.DataFrame Noramlized data with a mean of 0 and variance of 1 across the specified axis. """ # Normalize these values to range from 0 to 1 if axis == 1: standardized = data2d else: standardized = data2d.T subtract = standardized.min() standardized = (standardized - subtract) / ( standardized.max() - standardized.min()) if axis == 1: return standardized else: return standardized.T def dim_ratios(self, colors, dendrogram_ratio, colors_ratio): """Get the proportions of the figure taken up by each axes.""" ratios = [dendrogram_ratio] if colors is not None: # Colors are encoded as rgb, so there is an extra dimension if np.ndim(colors) > 2: n_colors = len(colors) else: n_colors = 1 ratios += [n_colors * colors_ratio] # Add the ratio for the heatmap itself ratios.append(1 - sum(ratios)) return ratios @staticmethod def color_list_to_matrix_and_cmap(colors, ind, axis=0): """Turns a list of colors into a numpy matrix and matplotlib colormap These arguments can now be plotted using heatmap(matrix, cmap) and the provided colors will be plotted. Parameters ---------- colors : list of matplotlib colors Colors to label the rows or columns of a dataframe. ind : list of ints Ordering of the rows or columns, to reorder the original colors by the clustered dendrogram order axis : int Which axis this is labeling Returns ------- matrix : numpy.array A numpy array of integer values, where each indexes into the cmap cmap : matplotlib.colors.ListedColormap """ try: mpl.colors.to_rgb(colors[0]) except ValueError: # We have a 2D color structure m, n = len(colors), len(colors[0]) if not all(len(c) == n for c in colors[1:]): raise ValueError("Multiple side color vectors must have same size") else: # We have one vector of colors m, n = 1, len(colors) colors = [colors] # Map from unique colors to colormap index value unique_colors = {} matrix = np.zeros((m, n), int) for i, inner in enumerate(colors): for j, color in enumerate(inner): idx = unique_colors.setdefault(color, len(unique_colors)) matrix[i, j] = idx # Reorder for clustering and transpose for axis matrix = matrix[:, ind] if axis == 0: matrix = matrix.T cmap = mpl.colors.ListedColormap(list(unique_colors)) return matrix, cmap def plot_dendrograms(self, row_cluster, col_cluster, metric, method, row_linkage, col_linkage, tree_kws): # Plot the row dendrogram if row_cluster: self.dendrogram_row = dendrogram( self.data2d, metric=metric, method=method, label=False, axis=0, ax=self.ax_row_dendrogram, rotate=True, linkage=row_linkage, tree_kws=tree_kws ) else: self.ax_row_dendrogram.set_xticks([]) self.ax_row_dendrogram.set_yticks([]) # PLot the column dendrogram if col_cluster: self.dendrogram_col = dendrogram( self.data2d, metric=metric, method=method, label=False, axis=1, ax=self.ax_col_dendrogram, linkage=col_linkage, tree_kws=tree_kws ) else: self.ax_col_dendrogram.set_xticks([]) self.ax_col_dendrogram.set_yticks([]) despine(ax=self.ax_row_dendrogram, bottom=True, left=True) despine(ax=self.ax_col_dendrogram, bottom=True, left=True) def plot_colors(self, xind, yind, **kws): """Plots color labels between the dendrogram and the heatmap Parameters ---------- heatmap_kws : dict Keyword arguments heatmap """ # Remove any custom colormap and centering # TODO this code has consistently caused problems when we # have missed kwargs that need to be excluded that it might # be better to rewrite *in*clusively. kws = kws.copy() kws.pop('cmap', None) kws.pop('norm', None) kws.pop('center', None) kws.pop('annot', None) kws.pop('vmin', None) kws.pop('vmax', None) kws.pop('robust', None) kws.pop('xticklabels', None) kws.pop('yticklabels', None) # Plot the row colors if self.row_colors is not None: matrix, cmap = self.color_list_to_matrix_and_cmap( self.row_colors, yind, axis=0) # Get row_color labels if self.row_color_labels is not None: row_color_labels = self.row_color_labels else: row_color_labels = False heatmap(matrix, cmap=cmap, cbar=False, ax=self.ax_row_colors, xticklabels=row_color_labels, yticklabels=False, **kws) # Adjust rotation of labels if row_color_labels is not False: plt.setp(self.ax_row_colors.get_xticklabels(), rotation=90) else: despine(self.ax_row_colors, left=True, bottom=True) # Plot the column colors if self.col_colors is not None: matrix, cmap = self.color_list_to_matrix_and_cmap( self.col_colors, xind, axis=1) # Get col_color labels if self.col_color_labels is not None: col_color_labels = self.col_color_labels else: col_color_labels = False heatmap(matrix, cmap=cmap, cbar=False, ax=self.ax_col_colors, xticklabels=False, yticklabels=col_color_labels, **kws) # Adjust rotation of labels, place on right side if col_color_labels is not False: self.ax_col_colors.yaxis.tick_right() plt.setp(self.ax_col_colors.get_yticklabels(), rotation=0) else: despine(self.ax_col_colors, left=True, bottom=True) def plot_matrix(self, colorbar_kws, xind, yind, **kws): self.data2d = self.data2d.iloc[yind, xind] self.mask = self.mask.iloc[yind, xind] # Try to reorganize specified tick labels, if provided xtl = kws.pop("xticklabels", "auto") try: xtl = np.asarray(xtl)[xind] except (TypeError, IndexError): pass ytl = kws.pop("yticklabels", "auto") try: ytl = np.asarray(ytl)[yind] except (TypeError, IndexError): pass # Reorganize the annotations to match the heatmap annot = kws.pop("annot", None) if annot is None or annot is False: pass else: if isinstance(annot, bool): annot_data = self.data2d else: annot_data = np.asarray(annot) if annot_data.shape != self.data2d.shape: err = "`data` and `annot` must have same shape." raise ValueError(err) annot_data = annot_data[yind][:, xind] annot = annot_data # Setting ax_cbar=None in clustermap call implies no colorbar kws.setdefault("cbar", self.ax_cbar is not None) heatmap(self.data2d, ax=self.ax_heatmap, cbar_ax=self.ax_cbar, cbar_kws=colorbar_kws, mask=self.mask, xticklabels=xtl, yticklabels=ytl, annot=annot, **kws) ytl = self.ax_heatmap.get_yticklabels() ytl_rot = None if not ytl else ytl[0].get_rotation() self.ax_heatmap.yaxis.set_ticks_position('right') self.ax_heatmap.yaxis.set_label_position('right') if ytl_rot is not None: ytl = self.ax_heatmap.get_yticklabels() plt.setp(ytl, rotation=ytl_rot) tight_params = dict(h_pad=.02, w_pad=.02) if self.ax_cbar is None: self._figure.tight_layout(**tight_params) else: # Turn the colorbar axes off for tight layout so that its # ticks don't interfere with the rest of the plot layout. # Then move it. self.ax_cbar.set_axis_off() self._figure.tight_layout(**tight_params) self.ax_cbar.set_axis_on() self.ax_cbar.set_position(self.cbar_pos) def plot(self, metric, method, colorbar_kws, row_cluster, col_cluster, row_linkage, col_linkage, tree_kws, **kws): # heatmap square=True sets the aspect ratio on the axes, but that is # not compatible with the multi-axes layout of clustergrid if kws.get("square", False): msg = "``square=True`` ignored in clustermap" warnings.warn(msg) kws.pop("square") colorbar_kws = {} if colorbar_kws is None else colorbar_kws self.plot_dendrograms(row_cluster, col_cluster, metric, method, row_linkage=row_linkage, col_linkage=col_linkage, tree_kws=tree_kws) try: xind = self.dendrogram_col.reordered_ind except AttributeError: xind = np.arange(self.data2d.shape[1]) try: yind = self.dendrogram_row.reordered_ind except AttributeError: yind = np.arange(self.data2d.shape[0]) self.plot_colors(xind, yind, **kws) self.plot_matrix(colorbar_kws, xind, yind, **kws) return self def clustermap( data, *, pivot_kws=None, method='average', metric='euclidean', z_score=None, standard_scale=None, figsize=(10, 10), cbar_kws=None, row_cluster=True, col_cluster=True, row_linkage=None, col_linkage=None, row_colors=None, col_colors=None, mask=None, dendrogram_ratio=.2, colors_ratio=0.03, cbar_pos=(.02, .8, .05, .18), tree_kws=None, **kwargs ): """ Plot a matrix dataset as a hierarchically-clustered heatmap. This function requires scipy to be available. Parameters ---------- data : 2D array-like Rectangular data for clustering. Cannot contain NAs. pivot_kws : dict, optional If `data` is a tidy dataframe, can provide keyword arguments for pivot to create a rectangular dataframe. method : str, optional Linkage method to use for calculating clusters. See :func:`scipy.cluster.hierarchy.linkage` documentation for more information. metric : str, optional Distance metric to use for the data. See :func:`scipy.spatial.distance.pdist` documentation for more options. To use different metrics (or methods) for rows and columns, you may construct each linkage matrix yourself and provide them as `{row,col}_linkage`. z_score : int or None, optional Either 0 (rows) or 1 (columns). Whether or not to calculate z-scores for the rows or the columns. Z scores are: z = (x - mean)/std, so values in each row (column) will get the mean of the row (column) subtracted, then divided by the standard deviation of the row (column). This ensures that each row (column) has mean of 0 and variance of 1. standard_scale : int or None, optional Either 0 (rows) or 1 (columns). Whether or not to standardize that dimension, meaning for each row or column, subtract the minimum and divide each by its maximum. figsize : tuple of (width, height), optional Overall size of the figure. cbar_kws : dict, optional Keyword arguments to pass to `cbar_kws` in :func:`heatmap`, e.g. to add a label to the colorbar. {row,col}_cluster : bool, optional If ``True``, cluster the {rows, columns}. {row,col}_linkage : :class:`numpy.ndarray`, optional Precomputed linkage matrix for the rows or columns. See :func:`scipy.cluster.hierarchy.linkage` for specific formats. {row,col}_colors : list-like or pandas DataFrame/Series, optional List of colors to label for either the rows or columns. Useful to evaluate whether samples within a group are clustered together. Can use nested lists or DataFrame for multiple color levels of labeling. If given as a :class:`pandas.DataFrame` or :class:`pandas.Series`, labels for the colors are extracted from the DataFrames column names or from the name of the Series. DataFrame/Series colors are also matched to the data by their index, ensuring colors are drawn in the correct order. mask : bool array or DataFrame, optional If passed, data will not be shown in cells where `mask` is True. Cells with missing values are automatically masked. Only used for visualizing, not for calculating. {dendrogram,colors}_ratio : float, or pair of floats, optional Proportion of the figure size devoted to the two marginal elements. If a pair is given, they correspond to (row, col) ratios. cbar_pos : tuple of (left, bottom, width, height), optional Position of the colorbar axes in the figure. Setting to ``None`` will disable the colorbar. tree_kws : dict, optional Parameters for the :class:`matplotlib.collections.LineCollection` that is used to plot the lines of the dendrogram tree. kwargs : other keyword arguments All other keyword arguments are passed to :func:`heatmap`. Returns ------- :class:`ClusterGrid` A :class:`ClusterGrid` instance. See Also -------- heatmap : Plot rectangular data as a color-encoded matrix. Notes ----- The returned object has a ``savefig`` method that should be used if you want to save the figure object without clipping the dendrograms. To access the reordered row indices, use: ``clustergrid.dendrogram_row.reordered_ind`` Column indices, use: ``clustergrid.dendrogram_col.reordered_ind`` Examples -------- .. include:: ../docstrings/clustermap.rst """ if _no_scipy: raise RuntimeError("clustermap requires scipy to be available") plotter = ClusterGrid(data, pivot_kws=pivot_kws, figsize=figsize, row_colors=row_colors, col_colors=col_colors, z_score=z_score, standard_scale=standard_scale, mask=mask, dendrogram_ratio=dendrogram_ratio, colors_ratio=colors_ratio, cbar_pos=cbar_pos) return plotter.plot(metric=metric, method=method, colorbar_kws=cbar_kws, row_cluster=row_cluster, col_cluster=col_cluster, row_linkage=row_linkage, col_linkage=col_linkage, tree_kws=tree_kws, **kwargs)