|
1
|
1 #!/usr/bin/env python
|
|
|
2
|
|
|
3 import os,sys,matplotlib,argparse,string
|
|
|
4 matplotlib.use('Agg')
|
|
|
5 from pylab import *
|
|
|
6 from lefse import *
|
|
|
7 import numpy as np
|
|
|
8
|
|
|
9 colors = ['r','g','b','m','c',[1.0,0.5,0.0],[0.0,1.0,0.0],[0.33,0.125,0.0],[0.75,0.75,0.75],'k']
|
|
|
10 dark_colors = [[0.4,0.0,0.0],[0.0,0.2,0.0],[0.0,0.0,0.4],'m','c',[1.0,0.5,0.0],[0.0,1.0,0.0],[0.33,0.125,0.0],[0.75,0.75,0.75],'k']
|
|
|
11
|
|
|
12 class CladeNode:
|
|
|
13 def __init__(self, name, abundance, viz = True):
|
|
|
14 self.id = name
|
|
|
15 self.name = name.split(".")
|
|
|
16 self.last_name = self.name[-1]
|
|
|
17 self.abundance = abundance
|
|
|
18 self.pos = (-1.0,-1.0)
|
|
|
19 self.children = {}
|
|
|
20 self.isleaf = True
|
|
|
21 self.color = 'y'
|
|
|
22 self.next_leaf = -1
|
|
|
23 self.prev_leaf = -1
|
|
|
24 self.viz = viz
|
|
|
25 def __repr__(self):
|
|
|
26 return self.last_name
|
|
|
27 def add_child(self,node):
|
|
|
28 self.isleaf = False
|
|
|
29 self.children[node.__repr__()] = node
|
|
|
30 def get_children(self):
|
|
|
31 ck = sorted(self.children.keys())
|
|
|
32 return [self.children[k] for k in ck]
|
|
|
33 def get_color(self):
|
|
|
34 return self.color
|
|
|
35 def set_color(self,c):
|
|
|
36 self.color = c
|
|
|
37 def set_pos(self,pos):
|
|
|
38 self.pos = pos
|
|
|
39
|
|
|
40 def read_params(args):
|
|
|
41 parser = argparse.ArgumentParser(description='Cladoplot')
|
|
|
42 parser.add_argument('input_file', metavar='INPUT_FILE', type=str, help="tab delimited input file")
|
|
|
43 parser.add_argument('output_file', metavar='OUTPUT_FILE', type=str, help="the file for the output image")
|
|
|
44 parser.add_argument('--clade_sep',dest="clade_sep", type=float, default=1.5)
|
|
|
45 parser.add_argument('--max_lev',dest="max_lev", type=int, default=-1)
|
|
|
46 parser.add_argument('--max_point_size',dest="max_point_size", type=float, default=6.0)
|
|
|
47 parser.add_argument('--min_point_size',dest="min_point_size", type=float, default=1)
|
|
|
48 parser.add_argument('--point_edge_width',dest="markeredgewidth", type=float, default=.25)
|
|
|
49 parser.add_argument('--siblings_connector_width',dest="siblings_connector_width", type=float, default=2)
|
|
|
50 parser.add_argument('--parents_connector_width',dest="parents_connector_width", type=float, default=0.75)
|
|
|
51 parser.add_argument('--radial_start_lev',dest="radial_start_lev", type=int, default=1)
|
|
|
52 parser.add_argument('--labeled_start_lev',dest="labeled_start_lev", type=int, default=2)
|
|
|
53 parser.add_argument('--labeled_stop_lev',dest="labeled_stop_lev", type=int, default=5)
|
|
|
54 parser.add_argument('--abrv_start_lev',dest="abrv_start_lev", type=int, default=3)
|
|
|
55 parser.add_argument('--abrv_stop_lev',dest="abrv_stop_lev", type=int, default=5)
|
|
|
56 parser.add_argument('--expand_void_lev',dest="expand_void_lev", type=int, default=1)
|
|
|
57 parser.add_argument('--class_legend_vis',dest="class_legend_vis", type=int, default=1)
|
|
|
58 parser.add_argument('--colored_connector',dest="colored_connectors", type=int, default=1)
|
|
|
59 parser.add_argument('--alpha',dest="alpha", type=float, default=0.2)
|
|
|
60 parser.add_argument('--title',dest="title", type=str, default="Cladogram")
|
|
|
61 parser.add_argument('--sub_clade',dest="sub_clade", type=str, default="")
|
|
|
62 parser.add_argument('--title_font_size',dest="title_font_size", type=str, default="14")
|
|
|
63 parser.add_argument('--right_space_prop',dest="r_prop", type=float, default=0.1)
|
|
|
64 parser.add_argument('--left_space_prop',dest="l_prop", type=float, default=0.1)
|
|
|
65 parser.add_argument('--label_font_size',dest="label_font_size", type=str, default="6")
|
|
|
66 parser.add_argument('--background_color',dest="back_color", type=str, choices=["k","w"], default="w", help="set the color of the background")
|
|
|
67 parser.add_argument('--colored_labels',dest="col_lab", type=int, choices=[0,1], default=1, help="draw the label with class color (1) or in black (0)")
|
|
|
68 parser.add_argument('--class_legend_font_size',dest="class_legend_font_size", type=str, default="10")
|
|
|
69 parser.add_argument('--dpi',dest="dpi", type=int, default=72)
|
|
|
70 parser.add_argument('--format', dest="format", choices=["png","svg","pdf"], default="svg", type=str, help="the format for the output file")
|
|
|
71 parser.add_argument('--all_feats', dest="all_feats", type=str, default="")
|
|
|
72 args = parser.parse_args()
|
|
|
73 return vars(args)
|
|
|
74
|
|
|
75 def cmp_names(la,lb):
|
|
|
76 if len(la) != len(lb): return False
|
|
|
77 for p in [(a,b) for i,a in enumerate(la) for j,b in enumerate(lb) if i == j]:
|
|
|
78 if p[0] != p[1]: return False
|
|
|
79 return True
|
|
|
80
|
|
|
81 def build_tree(father,all_nodes,l,depth,viz):
|
|
|
82 cc = [n for n in all_nodes if len(n.name) > len(father.name) and cmp_names(father.name,n.name[:len(father.name)])]
|
|
|
83 children = [n for n in cc if len(n.name) == len(father.name)+1]
|
|
|
84 if len(children) == 0 and l < depth -1: # !!!
|
|
|
85 nc = CladeNode(father.id+"."+father.id.split(".")[-1],1.0,viz)
|
|
|
86 father.add_child(nc)
|
|
|
87 children.append(nc)
|
|
|
88 for child in children:
|
|
|
89 build_tree(child,cc,l+1,depth,viz)
|
|
|
90 father.add_child(child)
|
|
|
91
|
|
|
92 def get_all_nodes(father):
|
|
|
93 ret = [father]
|
|
|
94 children = father.get_children()
|
|
|
95 for c in children:
|
|
|
96 ret += get_all_nodes(c)
|
|
|
97 return ret
|
|
|
98
|
|
|
99 def read_data(input_file,params):
|
|
|
100 with open(input_file, 'r') as inp:
|
|
|
101 if params['sub_clade'] == "": rows = [line.strip().split()[:-1] for line in inp.readlines() if params['max_lev'] < 1 or line.split()[0].count(".") < params['max_lev']]
|
|
|
102 else: rows = [line.split(params['sub_clade']+".")[1].strip().split()[:-1] for line in inp.readlines() if ( params['max_lev'] < 1 or line.split()[0].count(".") < params['max_lev'] ) and line.startswith(params['sub_clade']+".")]
|
|
|
103 all_names = [lin[0] for lin in rows]
|
|
|
104 to_add = []
|
|
|
105
|
|
|
106 abundances = [float(v) for v in zip(*rows)[1] if v >= 0.0]
|
|
|
107 tree = {}
|
|
|
108 tree['classes'] = list(set([v[2] for v in rows if len(v)>2]))
|
|
|
109 tree['classes'].sort()
|
|
|
110 all_nodes = [CladeNode("root."+row[0],float(row[1])) for row in rows]
|
|
|
111
|
|
|
112 depth = max([len(n.name) for n in all_nodes])
|
|
|
113
|
|
|
114 n2 = ["_".join(nn.name) for nn in all_nodes]
|
|
|
115 for i,nn in enumerate(all_nodes):
|
|
|
116 n = nn
|
|
|
117 while "_".join(n.name[:-1]) not in n2 and len(n.name) > 1:
|
|
|
118 n = CladeNode(".".join(n.name[:-1]),n.abundance)
|
|
|
119 all_nodes.append(n)
|
|
|
120 n2.append("_".join(n.name))
|
|
|
121
|
|
|
122 cls2 = []
|
|
|
123 if params['all_feats'] != "":
|
|
|
124 cls2 = sorted(params['all_feats'].split(":"))
|
|
|
125 for i,v in enumerate(rows):
|
|
|
126 if len(v)>2:
|
|
|
127 if len(cls2) > 0: all_nodes[i].set_color(colors[cls2.index(v[2])%len(colors)])
|
|
|
128 else:
|
|
|
129 if v[2].count('rgbcol') > 0:
|
|
|
130 ccc = [float(tt) for tt in v[2].split('_')[1:]]
|
|
|
131 all_nodes[i].set_color(ccc)
|
|
|
132 else: all_nodes[i].set_color(colors[sorted(tree['classes']).index(v[2])%len(colors)])
|
|
|
133 root = CladeNode("root",-1.0)
|
|
|
134 root.set_pos((0.0,0.0))
|
|
|
135
|
|
|
136 build_tree(root,all_nodes,0,depth,params['expand_void_lev']==1)
|
|
|
137
|
|
|
138 all_nodes = get_all_nodes(root)
|
|
|
139
|
|
|
140 tree['root'] = root
|
|
|
141 tree['max_abs'] = max(abundances)
|
|
|
142 tree['min_abs'] = min(abundances)
|
|
|
143 levs = []
|
|
|
144 for i in range(depth):
|
|
|
145 depthi = [n for n in all_nodes if len(n.name) == i+1]
|
|
|
146 levs.append(len(depthi))
|
|
|
147 tree['nlev'] = levs
|
|
|
148 return tree
|
|
|
149
|
|
|
150 def add_all_pos(father,n,distn,seps,tsep,mlev,last_leaf=-1,nc=1):
|
|
|
151 children = father.get_children()
|
|
|
152 leaves = True if children[0].isleaf else False
|
|
|
153 for i,child in enumerate(children):
|
|
|
154 if leaves:
|
|
|
155 n += 1.0
|
|
|
156 men = 0.5 if len(children) == 1 else 0.0
|
|
|
157 child.set_pos((n*distn-men*float(distn)+tsep,(len(father.name))/float(mlev-1)))
|
|
|
158 if last_leaf != -1:
|
|
|
159 child.prev_leaf = last_leaf
|
|
|
160 last_leaf.next_leaf = child
|
|
|
161 last_leaf = child
|
|
|
162 else:
|
|
|
163 ln = n
|
|
|
164 ltsep = tsep
|
|
|
165 n,tsep,last_leaf = add_all_pos(child,n,distn,seps,tsep,mlev,last_leaf,len(children))
|
|
|
166 nn = (ln + n)*0.5*distn
|
|
|
167 ssep = (ltsep + tsep)*0.5
|
|
|
168 if n-ln == 1:
|
|
|
169 ssep = ltsep
|
|
|
170 child.set_pos((nn+ssep,(len(father.name))/float(mlev-1)))
|
|
|
171 tsep += seps[len(father.name)-1]
|
|
|
172 return n,tsep,last_leaf
|
|
|
173
|
|
|
174 def plot_points(father,params,pt_scale,ax):
|
|
|
175 children = father.get_children()
|
|
|
176 children.sort(key = lambda a: -int(a.get_color() == 'y')*a.abundance)
|
|
|
177 x,r = father.pos[0], father.pos[1]
|
|
|
178 for i,child in enumerate(children):
|
|
|
179 xc,rc = plot_points(child,params,pt_scale,ax)
|
|
|
180 if not father.viz: return x,r
|
|
|
181 ps = pt_scale[0]+father.abundance/pt_scale[1]+pt_scale[0]
|
|
|
182 col = father.get_color()
|
|
|
183 pw = params['markeredgewidth'] if col == 'y' else params['markeredgewidth']*3.0
|
|
|
184 if x==0 and r==0: ax.plot(x,r, 'o',markersize=ps,color=col,markeredgewidth=0.01,markeredgecolor=params['fore_color'])
|
|
|
185 else: ax.plot(x,r, 'o',markersize=ps,color=col,markeredgewidth=pw,markeredgecolor=params['fore_color'])
|
|
|
186 return x,r
|
|
|
187
|
|
|
188 def plot_lines(father,params,depth,ax,xf):
|
|
|
189 children = father.get_children()
|
|
|
190 x,r = father.pos[0], father.pos[1]
|
|
|
191 for i,child in enumerate(children):
|
|
|
192 xc,rc = plot_lines(child,params,depth,ax,x)
|
|
|
193 if i == 0: x_first, r_first = xc, rc
|
|
|
194 if len(father.name) >= depth-params['radial_start_lev']:
|
|
|
195 col = params['fore_color']
|
|
|
196 lw=params['parents_connector_width']
|
|
|
197 if not child.viz: continue
|
|
|
198 if father.get_color() != 'y' and father.get_color() == child.get_color() and params['colored_connectors']:
|
|
|
199 col = child.get_color()
|
|
|
200 lw *=2.5
|
|
|
201 if col != params['fore_color']:
|
|
|
202 ax.plot([x,xc],[r,rc],"-",color=params['fore_color'],lw=lw*1.5)
|
|
|
203 ax.plot([x,xc],[r,rc],"-",color=col,lw=lw)
|
|
|
204
|
|
|
205 if not father.viz or (len(children) == 1 and not children[0].viz): return x,r
|
|
|
206 if len(father.name) < depth-params['radial_start_lev']:
|
|
|
207 col = params['fore_color']
|
|
|
208 lw=params['parents_connector_width']
|
|
|
209 if father.get_color() != 'y':
|
|
|
210 f =True
|
|
|
211 for child in children:
|
|
|
212 if child.get_color() != father.get_color() or not params['colored_connectors']:
|
|
|
213 f = False
|
|
|
214 break
|
|
|
215 if f:
|
|
|
216 col = father.get_color()
|
|
|
217 lw *= 2.5
|
|
|
218 if not (x==0 and r==0):
|
|
|
219 xx = xc if len(children) > 0 else x
|
|
|
220 if len(children) == 0: rc = r
|
|
|
221 xt = x if len(children)>1 else xx
|
|
|
222 if col != params['fore_color']:
|
|
|
223 ax.plot([x,xt],[r,rc],"-",color=params['fore_color'],lw=lw*1.5)
|
|
|
224 ax.plot([x,xt],[r,rc],"-",color=col,lw=lw)
|
|
|
225 if len(children) > 0 and 1 < len(father.name) < depth-params['radial_start_lev']:
|
|
|
226 xs = arange(x_first,xc,0.01)
|
|
|
227 ys = [rc for t in xs]
|
|
|
228 ax.plot(xs,ys,"-",color=col,lw=params['siblings_connector_width'],markeredgecolor=params['fore_color'])
|
|
|
229 return x,r
|
|
|
230
|
|
|
231 def uniqueid():
|
|
|
232 for l in string.lowercase: yield l
|
|
|
233 for l in string.lowercase:
|
|
|
234 for i in range(10):
|
|
|
235 yield l+str(i)
|
|
|
236 i = 0
|
|
|
237 while True:
|
|
|
238 yield str(i)
|
|
|
239 i += 1
|
|
|
240
|
|
|
241 def plot_names(father,params,depth,ax,u_i,seps):
|
|
|
242 children = father.get_children()
|
|
|
243 l = len(father.name)
|
|
|
244 if len(children)==0:
|
|
|
245 if father.prev_leaf == -1 or father.next_leaf == -1:
|
|
|
246 fr_0, fr_1 = father.pos[0], father.pos[0]
|
|
|
247 else: fr_0, fr_1 = (father.pos[0]+father.prev_leaf.pos[0])*0.5, (father.pos[0]+father.next_leaf.pos[0])*0.5
|
|
|
248 for i,child in enumerate(children):
|
|
|
249 fr,to = plot_names(child,params,depth,ax,u_i,seps)
|
|
|
250 if i == 0: fr_0 = fr
|
|
|
251 fr_1 = to
|
|
|
252 if father.get_color() != 'y' and params['labeled_start_lev'] < l <= params['labeled_stop_lev']+1:
|
|
|
253 col = father.get_color()
|
|
|
254 dd = params['labeled_stop_lev'] - params['labeled_start_lev'] + 1
|
|
|
255 de = depth - 1
|
|
|
256 dim = 1.0/float(de)
|
|
|
257 perc_ext = 0.65 if dim > 0.1 else 1.0
|
|
|
258 clto = (de-l+1)*dim+dim*(dd+1-(l-dd-1))*perc_ext
|
|
|
259 clto = (de-l+1)*dim+dim*(dd-(l-params['labeled_start_lev'])+1)*perc_ext
|
|
|
260 des = float(180.0*(fr_0+fr_1)/np.pi)*0.5-90
|
|
|
261 lab = ""
|
|
|
262 txt = father.last_name
|
|
|
263 if params['abrv_start_lev'] < l <= params['abrv_stop_lev'] + 1:
|
|
|
264 ide = u_i.next()
|
|
|
265 lab = str(ide)+": "+father.last_name
|
|
|
266 txt = str(ide)
|
|
|
267 # ax.bar(fr_0, clto, width = fr_1-fr_0, bottom = float(l-1)/float(depth-1), alpha = params['alpha'], color=col, edgecolor=col)
|
|
|
268 ax.bar(fr_0, clto, width = fr_1-fr_0, bottom = float(l-1)/float(de), alpha = params['alpha'], color=col, edgecolor=col)
|
|
|
269 ax.bar(0.0, 0.0, width = 0.0, bottom = 0.0, alpha = 1.0, color=col, edgecolor=params['fore_color'], label=lab)
|
|
|
270 if l <= params['abrv_stop_lev'] + 1:
|
|
|
271 if not params['col_lab']: col = params['fore_color']
|
|
|
272 else:
|
|
|
273 if col not in colors: col = params['fore_color']
|
|
|
274 else: col = dark_colors[colors.index(col)%len(dark_colors)]
|
|
|
275 ax.text((fr_0+fr_1)*0.5, clto+float(l-1)/float(de)-dim*perc_ext/2.0, txt, size = params['label_font_size'], rotation=des, ha ="center", va="center", color=col)
|
|
|
276 return fr_0, fr_1
|
|
|
277
|
|
|
278 def draw_tree(out_file,tree,params):
|
|
|
279 plt_size = 7
|
|
|
280 nlev = tree['nlev']
|
|
|
281 pt_scale = (params['min_point_size'],max(1.0,((tree['max_abs']-tree['min_abs']))/(params['max_point_size']-params['min_point_size'])))
|
|
|
282 depth = len(nlev)
|
|
|
283 sep = (2.0*np.pi)/float(nlev[-1])
|
|
|
284 seps = [params['clade_sep']*sep/float(depth-i+1) for i in range(1,len(tree['nlev'])+1)]
|
|
|
285 totseps = sum([s*nlev[i] for i,s in enumerate(seps[:-1])])
|
|
|
286 clade_sep_err = True if totseps > np.pi else False
|
|
|
287 while totseps > np.pi:
|
|
|
288 params['clade_sep'] *= 0.75
|
|
|
289 seps = [params['clade_sep']*sep/(float(depth-i+1)*0.25) for i in range(1,len(tree['nlev'])+1)]
|
|
|
290 totseps = sum([s*nlev[i] for i,s in enumerate(seps[:-1])])
|
|
|
291 if clade_sep_err: print 'clade_sep parameter too large, lowered to',params['clade_sep']
|
|
|
292
|
|
|
293 fig = plt.figure(edgecolor=params['back_color'],facecolor=params['back_color'])
|
|
|
294 ax = fig.add_subplot(111, polar=True, frame_on=False, axis_bgcolor=params['back_color'] )
|
|
|
295 plt.subplots_adjust(right=1.0-params['r_prop'],left=params['l_prop'])
|
|
|
296 ax.grid(False)
|
|
|
297 xticks([])
|
|
|
298 yticks([])
|
|
|
299
|
|
|
300 ds = (2.0*np.pi-totseps)/float(nlev[-1])
|
|
|
301
|
|
|
302 add_all_pos(tree['root'],0.0,ds,seps,0.0,depth)
|
|
|
303
|
|
|
304 plot_lines(tree['root'],params,depth,ax,0)
|
|
|
305 plot_points(tree['root'],params,pt_scale,ax)
|
|
|
306 plot_names(tree['root'],params,depth,ax,uniqueid(),seps)
|
|
|
307
|
|
|
308 r = np.arange(0, 3.0, 0.01)
|
|
|
309 theta = 2*np.pi*r
|
|
|
310
|
|
|
311 def get_col_attr(x):
|
|
|
312 return hasattr(x, 'set_color') and not hasattr(x, 'set_facecolor')
|
|
|
313
|
|
|
314 h, l = ax.get_legend_handles_labels()
|
|
|
315 if len(l) > 0:
|
|
|
316 leg = ax.legend(bbox_to_anchor=(1.05, 1), frameon=False, loc=2, borderaxespad=0.,prop={'size':params['label_font_size']})
|
|
|
317 if leg != None:
|
|
|
318 gca().add_artist(leg)
|
|
|
319 for o in leg.findobj(get_col_attr):
|
|
|
320 o.set_color(params['fore_color'])
|
|
|
321
|
|
|
322 cll = sorted(tree['classes']) if params['all_feats'] == "" else sorted(params['all_feats'].split(":"))
|
|
|
323 nll = [ax.bar(0.0, 0.0, width = 0.0, bottom = 0.0, color=colors[i%len(colors)], label=c) for i,c in enumerate(cll) if c in tree['classes']]
|
|
|
324 cl = [c for c in cll if c in tree['classes']]
|
|
|
325
|
|
|
326 ax.set_title(params['title'],size=params['title_font_size'],color=params['fore_color'])
|
|
|
327
|
|
|
328 if params['class_legend_vis']:
|
|
|
329 l2 = legend(nll, cl, loc=2, prop={'size':params['class_legend_font_size']}, frameon=False)
|
|
|
330 if l2 != None:
|
|
|
331 for o in l2.findobj(get_col_attr):
|
|
|
332 o.set_color(params['fore_color'])
|
|
|
333
|
|
|
334 plt.savefig(out_file,format=params['format'],facecolor=params['back_color'],edgecolor=params['fore_color'],dpi=params['dpi'])
|
|
|
335 plt.close()
|
|
|
336
|
|
|
337 if __name__ == '__main__':
|
|
|
338 params = read_params(sys.argv)
|
|
|
339 params['fore_color'] = 'w' if params['back_color'] == 'k' else 'k'
|
|
|
340 clad_tree = read_data(params['input_file'],params)
|
|
|
341 draw_tree(params['output_file'],clad_tree,params)
|
|
|
342
|
