Package Bio :: Package Phylo :: Module _utils
[hide private]
[frames] | no frames]

Source Code for Module Bio.Phylo._utils

  1  # Copyright (C) 2009 by Eric Talevich (eric.talevich@gmail.com) 
  2  # This code is part of the Biopython distribution and governed by its 
  3  # license. Please see the LICENSE file that should have been included 
  4  # as part of this package. 
  5   
  6  """Utilities for handling, displaying and exporting Phylo trees. 
  7   
  8  Third-party libraries are loaded when the corresponding function is called. 
  9  """ 
 10  __docformat__ = "restructuredtext en" 
 11   
 12  import math 
 13  import sys 
 14   
 15   
16 -def to_networkx(tree):
17 """Convert a Tree object to a networkx graph. 18 19 The result is useful for graph-oriented analysis, and also interactive 20 plotting with pylab, matplotlib or pygraphviz, though the resulting diagram 21 is usually not ideal for displaying a phylogeny. 22 23 Requires NetworkX version 0.99 or later. 24 """ 25 try: 26 import networkx 27 except ImportError: 28 from Bio import MissingPythonDependencyError 29 raise MissingPythonDependencyError( 30 "Install NetworkX if you want to use to_networkx.") 31 32 # NB (1/2010): the networkx API stabilized at v.1.0 33 # 1.0+: edges accept arbitrary data as kwargs, weights are floats 34 # 0.99: edges accept weight as a string, nothing else 35 # pre-0.99: edges accept no additional data 36 # Ubuntu Lucid LTS uses v0.99, let's support everything 37 if networkx.__version__ >= '1.0': 38 def add_edge(graph, n1, n2): 39 graph.add_edge(n1, n2, weight=n2.branch_length or 1.0) 40 # Copy branch color value as hex, if available 41 if hasattr(n2, 'color') and n2.color is not None: 42 graph[n1][n2]['color'] = n2.color.to_hex() 43 elif hasattr(n1, 'color') and n1.color is not None: 44 # Cascading color attributes 45 graph[n1][n2]['color'] = n1.color.to_hex() 46 n2.color = n1.color 47 # Copy branch weight value (float) if available 48 if hasattr(n2, 'width') and n2.width is not None: 49 graph[n1][n2]['width'] = n2.width 50 elif hasattr(n1, 'width') and n1.width is not None: 51 # Cascading width attributes 52 graph[n1][n2]['width'] = n1.width 53 n2.width = n1.width
54 elif networkx.__version__ >= '0.99': 55 def add_edge(graph, n1, n2): 56 graph.add_edge(n1, n2, (n2.branch_length or 1.0)) 57 else: 58 def add_edge(graph, n1, n2): 59 graph.add_edge(n1, n2) 60 61 def build_subgraph(graph, top): 62 """Walk down the Tree, building graphs, edges and nodes.""" 63 for clade in top: 64 graph.add_node(clade.root) 65 add_edge(graph, top.root, clade.root) 66 build_subgraph(graph, clade) 67 68 if tree.rooted: 69 G = networkx.DiGraph() 70 else: 71 G = networkx.Graph() 72 G.add_node(tree.root) 73 build_subgraph(G, tree.root) 74 return G 75 76
77 -def draw_graphviz(tree, label_func=str, prog='twopi', args='', 78 node_color='#c0deff', **kwargs):
79 """Display a tree or clade as a graph, using the graphviz engine. 80 81 Requires NetworkX, matplotlib, Graphviz and either PyGraphviz or pydot. 82 83 The third and fourth parameters apply to Graphviz, and the remaining 84 arbitrary keyword arguments are passed directly to networkx.draw(), which 85 in turn mostly wraps matplotlib/pylab. See the documentation for Graphviz 86 and networkx for detailed explanations. 87 88 The NetworkX/matplotlib parameters are described in the docstrings for 89 networkx.draw() and pylab.scatter(), but the most reasonable options to try 90 are: *alpha, node_color, node_size, node_shape, edge_color, style, 91 font_size, font_color, font_weight, font_family* 92 93 :Parameters: 94 95 label_func : callable 96 A function to extract a label from a node. By default this is str(), 97 but you can use a different function to select another string 98 associated with each node. If this function returns None for a node, 99 no label will be shown for that node. 100 101 The label will also be silently skipped if the throws an exception 102 related to ordinary attribute access (LookupError, AttributeError, 103 ValueError); all other exception types will still be raised. This 104 means you can use a lambda expression that simply attempts to look 105 up the desired value without checking if the intermediate attributes 106 are available: 107 108 >>> Phylo.draw_graphviz(tree, lambda n: n.taxonomies[0].code) 109 110 prog : string 111 The Graphviz program to use when rendering the graph. 'twopi' 112 behaves the best for large graphs, reliably avoiding crossing edges, 113 but for moderate graphs 'neato' looks a bit nicer. For small 114 directed graphs, 'dot' may produce a normal-looking cladogram, but 115 will cross and distort edges in larger graphs. (The programs 'circo' 116 and 'fdp' are not recommended.) 117 args : string 118 Options passed to the external graphviz program. Normally not 119 needed, but offered here for completeness. 120 121 Example 122 ------- 123 124 >>> import pylab 125 >>> from Bio import Phylo 126 >>> tree = Phylo.read('ex/apaf.xml', 'phyloxml') 127 >>> Phylo.draw_graphviz(tree) 128 >>> pylab.show() 129 >>> pylab.savefig('apaf.png') 130 """ 131 try: 132 import networkx 133 except ImportError: 134 from Bio import MissingPythonDependencyError 135 raise MissingPythonDependencyError( 136 "Install NetworkX if you want to use to_networkx.") 137 138 G = to_networkx(tree) 139 try: 140 # NetworkX version 1.8 or later (2013-01-20) 141 Gi = networkx.convert_node_labels_to_integers(G, 142 label_attribute='label') 143 int_labels = {} 144 for integer, nodeattrs in Gi.node.items(): 145 int_labels[nodeattrs['label']] = integer 146 except TypeError: 147 # Older NetworkX versions (before 1.8) 148 Gi = networkx.convert_node_labels_to_integers(G, 149 discard_old_labels=False) 150 int_labels = Gi.node_labels 151 152 try: 153 posi = networkx.graphviz_layout(Gi, prog, args=args) 154 except ImportError: 155 raise MissingPythonDependencyError( 156 "Install PyGraphviz or pydot if you want to use draw_graphviz.") 157 158 def get_label_mapping(G, selection): 159 """Apply the user-specified node relabeling.""" 160 for node in G.nodes(): 161 if (selection is None) or (node in selection): 162 try: 163 label = label_func(node) 164 if label not in (None, node.__class__.__name__): 165 yield (node, label) 166 except (LookupError, AttributeError, ValueError): 167 pass
168 169 if 'nodelist' in kwargs: 170 labels = dict(get_label_mapping(G, set(kwargs['nodelist']))) 171 else: 172 labels = dict(get_label_mapping(G, None)) 173 kwargs['nodelist'] = list(labels.keys()) 174 if 'edge_color' not in kwargs: 175 kwargs['edge_color'] = [isinstance(e[2], dict) and 176 e[2].get('color', 'k') or 'k' 177 for e in G.edges(data=True)] 178 if 'width' not in kwargs: 179 kwargs['width'] = [isinstance(e[2], dict) and 180 e[2].get('width', 1.0) or 1.0 181 for e in G.edges(data=True)] 182 183 posn = dict((n, posi[int_labels[n]]) for n in G) 184 networkx.draw(G, posn, labels=labels, node_color=node_color, **kwargs) 185 186
187 -def draw_ascii(tree, file=None, column_width=80):
188 """Draw an ascii-art phylogram of the given tree. 189 190 The printed result looks like:: 191 192 _________ Orange 193 ______________| 194 | |______________ Tangerine 195 ______________| 196 | | _________________________ Grapefruit 197 _| |_________| 198 | |______________ Pummelo 199 | 200 |__________________________________ Apple 201 202 203 :Parameters: 204 file : file-like object 205 File handle opened for writing the output drawing. (Default: 206 standard output) 207 column_width : int 208 Total number of text columns used by the drawing. 209 """ 210 if file is None: 211 file = sys.stdout 212 213 taxa = tree.get_terminals() 214 # Some constants for the drawing calculations 215 max_label_width = max(len(str(taxon)) for taxon in taxa) 216 drawing_width = column_width - max_label_width - 1 217 drawing_height = 2 * len(taxa) - 1 218 219 def get_col_positions(tree): 220 """Create a mapping of each clade to its column position.""" 221 depths = tree.depths() 222 # If there are no branch lengths, assume unit branch lengths 223 if not max(depths.values()): 224 depths = tree.depths(unit_branch_lengths=True) 225 # Potential drawing overflow due to rounding -- 1 char per tree layer 226 fudge_margin = int(math.ceil(math.log(len(taxa), 2))) 227 cols_per_branch_unit = ((drawing_width - fudge_margin) 228 / float(max(depths.values()))) 229 return dict((clade, int(blen * cols_per_branch_unit + 1.0)) 230 for clade, blen in depths.items())
231 232 def get_row_positions(tree): 233 positions = dict((taxon, 2 * idx) for idx, taxon in enumerate(taxa)) 234 235 def calc_row(clade): 236 for subclade in clade: 237 if subclade not in positions: 238 calc_row(subclade) 239 positions[clade] = ((positions[clade.clades[0]] + 240 positions[clade.clades[-1]]) // 2) 241 242 calc_row(tree.root) 243 return positions 244 245 col_positions = get_col_positions(tree) 246 row_positions = get_row_positions(tree) 247 char_matrix = [[' ' for x in range(drawing_width)] 248 for y in range(drawing_height)] 249 250 def draw_clade(clade, startcol): 251 thiscol = col_positions[clade] 252 thisrow = row_positions[clade] 253 # Draw a horizontal line 254 for col in range(startcol, thiscol): 255 char_matrix[thisrow][col] = '_' 256 if clade.clades: 257 # Draw a vertical line 258 toprow = row_positions[clade.clades[0]] 259 botrow = row_positions[clade.clades[-1]] 260 for row in range(toprow + 1, botrow + 1): 261 char_matrix[row][thiscol] = '|' 262 # NB: Short terminal branches need something to stop rstrip() 263 if (col_positions[clade.clades[0]] - thiscol) < 2: 264 char_matrix[toprow][thiscol] = ',' 265 # Draw descendents 266 for child in clade: 267 draw_clade(child, thiscol + 1) 268 269 draw_clade(tree.root, 0) 270 # Print the complete drawing 271 for idx, row in enumerate(char_matrix): 272 line = ''.join(row).rstrip() 273 # Add labels for terminal taxa in the right margin 274 if idx % 2 == 0: 275 line += ' ' + str(taxa[idx // 2]) 276 file.write(line + '\n') 277 file.write('\n') 278 279
280 -def draw(tree, label_func=str, do_show=True, show_confidence=True, 281 # For power users 282 axes=None, branch_labels=None, *args, **kwargs):
283 """Plot the given tree using matplotlib (or pylab). 284 285 The graphic is a rooted tree, drawn with roughly the same algorithm as 286 draw_ascii. 287 288 Additional keyword arguments passed into this function are used as pyplot 289 options. The input format should be in the form of: 290 pyplot_option_name=(tuple), pyplot_option_name=(tuple, dict), or 291 pyplot_option_name=(dict). 292 293 Example using the pyplot options 'axhspan' and 'axvline': 294 295 >>> Phylo.draw(tree, axhspan=((0.25, 7.75), {'facecolor':'0.5'}), 296 ... axvline={'x':'0', 'ymin':'0', 'ymax':'1'}) 297 298 Visual aspects of the plot can also be modified using pyplot's own functions 299 and objects (via pylab or matplotlib). In particular, the pyplot.rcParams 300 object can be used to scale the font size (rcParams["font.size"]) and line 301 width (rcParams["lines.linewidth"]). 302 303 :Parameters: 304 label_func : callable 305 A function to extract a label from a node. By default this is str(), 306 but you can use a different function to select another string 307 associated with each node. If this function returns None for a node, 308 no label will be shown for that node. 309 do_show : bool 310 Whether to show() the plot automatically. 311 show_confidence : bool 312 Whether to display confidence values, if present on the tree. 313 axes : matplotlib/pylab axes 314 If a valid matplotlib.axes.Axes instance, the phylogram is plotted 315 in that Axes. By default (None), a new figure is created. 316 branch_labels : dict or callable 317 A mapping of each clade to the label that will be shown along the 318 branch leading to it. By default this is the confidence value(s) of 319 the clade, taken from the ``confidence`` attribute, and can be 320 easily toggled off with this function's ``show_confidence`` option. 321 But if you would like to alter the formatting of confidence values, 322 or label the branches with something other than confidence, then use 323 this option. 324 """ 325 326 try: 327 import matplotlib.pyplot as plt 328 except ImportError: 329 try: 330 import pylab as plt 331 except ImportError: 332 from Bio import MissingPythonDependencyError 333 raise MissingPythonDependencyError( 334 "Install matplotlib or pylab if you want to use draw.") 335 336 import matplotlib.collections as mpcollections 337 338 # Arrays that store lines for the plot of clades 339 horizontal_linecollections = [] 340 vertical_linecollections = [] 341 342 # Options for displaying branch labels / confidence 343 def conf2str(conf): 344 if int(conf) == conf: 345 return str(int(conf)) 346 return str(conf)
347 if not branch_labels: 348 if show_confidence: 349 def format_branch_label(clade): 350 if hasattr(clade, 'confidences'): 351 # phyloXML supports multiple confidences 352 return '/'.join(conf2str(cnf.value) 353 for cnf in clade.confidences) 354 if clade.confidence: 355 return conf2str(clade.confidence) 356 return None 357 else: 358 def format_branch_label(clade): 359 return None 360 elif isinstance(branch_labels, dict): 361 def format_branch_label(clade): 362 return branch_labels.get(clade) 363 else: 364 assert callable(branch_labels), \ 365 "branch_labels must be either a dict or a callable (function)" 366 format_branch_label = branch_labels 367 368 # Layout 369 370 def get_x_positions(tree): 371 """Create a mapping of each clade to its horizontal position. 372 373 Dict of {clade: x-coord} 374 """ 375 depths = tree.depths() 376 # If there are no branch lengths, assume unit branch lengths 377 if not max(depths.values()): 378 depths = tree.depths(unit_branch_lengths=True) 379 return depths 380 381 def get_y_positions(tree): 382 """Create a mapping of each clade to its vertical position. 383 384 Dict of {clade: y-coord}. 385 Coordinates are negative, and integers for tips. 386 """ 387 maxheight = tree.count_terminals() 388 # Rows are defined by the tips 389 heights = dict((tip, maxheight - i) 390 for i, tip in enumerate(reversed(tree.get_terminals()))) 391 392 # Internal nodes: place at midpoint of children 393 def calc_row(clade): 394 for subclade in clade: 395 if subclade not in heights: 396 calc_row(subclade) 397 # Closure over heights 398 heights[clade] = (heights[clade.clades[0]] + 399 heights[clade.clades[-1]]) / 2.0 400 401 if tree.root.clades: 402 calc_row(tree.root) 403 return heights 404 405 x_posns = get_x_positions(tree) 406 y_posns = get_y_positions(tree) 407 # The function draw_clade closes over the axes object 408 if axes is None: 409 fig = plt.figure() 410 axes = fig.add_subplot(1, 1, 1) 411 elif not isinstance(axes, plt.matplotlib.axes.Axes): 412 raise ValueError("Invalid argument for axes: %s" % axes) 413 414 def draw_clade_lines(use_linecollection=False, orientation='horizontal', 415 y_here=0, x_start=0, x_here=0, y_bot=0, y_top=0, 416 color='black', lw='.1'): 417 """Create a line with or without a line collection object. 418 419 Graphical formatting of the lines representing clades in the plot can be 420 customized by altering this function. 421 """ 422 if (use_linecollection is False and orientation == 'horizontal'): 423 axes.hlines(y_here, x_start, x_here, color=color, lw=lw) 424 elif (use_linecollection is True and orientation == 'horizontal'): 425 horizontal_linecollections.append(mpcollections.LineCollection( 426 [[(x_start, y_here), (x_here, y_here)]], color=color, lw=lw),) 427 elif (use_linecollection is False and orientation == 'vertical'): 428 axes.vlines(x_here, y_bot, y_top, color=color) 429 elif (use_linecollection is True and orientation == 'vertical'): 430 vertical_linecollections.append(mpcollections.LineCollection( 431 [[(x_here, y_bot), (x_here, y_top)]], color=color, lw=lw),) 432 433 def draw_clade(clade, x_start, color, lw): 434 """Recursively draw a tree, down from the given clade.""" 435 x_here = x_posns[clade] 436 y_here = y_posns[clade] 437 # phyloXML-only graphics annotations 438 if hasattr(clade, 'color') and clade.color is not None: 439 color = clade.color.to_hex() 440 if hasattr(clade, 'width') and clade.width is not None: 441 lw = clade.width * plt.rcParams['lines.linewidth'] 442 # Draw a horizontal line from start to here 443 draw_clade_lines(use_linecollection=True, orientation='horizontal', 444 y_here=y_here, x_start=x_start, x_here=x_here, color=color, lw=lw) 445 # Add node/taxon labels 446 label = label_func(clade) 447 if label not in (None, clade.__class__.__name__): 448 axes.text(x_here, y_here, ' %s' % 449 label, verticalalignment='center') 450 # Add label above the branch (optional) 451 conf_label = format_branch_label(clade) 452 if conf_label: 453 axes.text(0.5 * (x_start + x_here), y_here, conf_label, 454 fontsize='small', horizontalalignment='center') 455 if clade.clades: 456 # Draw a vertical line connecting all children 457 y_top = y_posns[clade.clades[0]] 458 y_bot = y_posns[clade.clades[-1]] 459 # Only apply widths to horizontal lines, like Archaeopteryx 460 draw_clade_lines(use_linecollection=True, orientation='vertical', 461 x_here=x_here, y_bot=y_bot, y_top=y_top, color=color, lw=lw) 462 # Draw descendents 463 for child in clade: 464 draw_clade(child, x_here, color, lw) 465 466 draw_clade(tree.root, 0, 'k', plt.rcParams['lines.linewidth']) 467 468 # If line collections were used to create clade lines, here they are added 469 # to the pyplot plot. 470 for i in horizontal_linecollections: 471 axes.add_collection(i) 472 for i in vertical_linecollections: 473 axes.add_collection(i) 474 475 # Aesthetics 476 477 if hasattr(tree, 'name') and tree.name: 478 axes.set_title(tree.name) 479 axes.set_xlabel('branch length') 480 axes.set_ylabel('taxa') 481 # Add margins around the tree to prevent overlapping the axes 482 xmax = max(x_posns.values()) 483 axes.set_xlim(-0.05 * xmax, 1.25 * xmax) 484 # Also invert the y-axis (origin at the top) 485 # Add a small vertical margin, but avoid including 0 and N+1 on the y axis 486 axes.set_ylim(max(y_posns.values()) + 0.8, 0.2) 487 488 # Parse and process key word arguments as pyplot options 489 for key, value in kwargs.items(): 490 try: 491 # Check that the pyplot option input is iterable, as required 492 [i for i in value] 493 except TypeError: 494 raise ValueError('Keyword argument "%s=%s" is not in the format ' 495 'pyplot_option_name=(tuple), pyplot_option_name=(tuple, dict),' 496 ' or pyplot_option_name=(dict) ' 497 % (key, value)) 498 if isinstance(value, dict): 499 getattr(plt, str(key))(**dict(value)) 500 elif not (isinstance(value[0], tuple)): 501 getattr(plt, str(key))(*value) 502 elif (isinstance(value[0], tuple)): 503 getattr(plt, str(key))(*value[0], **dict(value[1])) 504 505 if do_show: 506 plt.show() 507