source: mds-and-trees/tree-genealogy.py @ 619

Last change on this file since 619 was 619, checked in by konrad, 8 years ago

Changed spine and skeleton drawing methods from recurrent to iterative ones.

File size: 19.1 KB
Line 
1# Draws a genealogical tree (generates a SVG file) based on parent-child relationship information.
2# Supports files generated by Framsticks experiments.
3
4import json
5import random
6import math
7import argparse
8
9TIME = "" # BIRTHS / GENERATIONAL / REAL
10BALANCE = "" # MIN / DENSITY
11
12DOT_STYLE = "" # NONE / NORMAL / CLEAR
13
14JITTER = "" #
15
16# ------SVG---------
17svg_file = 0
18
19svg_line_style = 'stroke="rgb(90%,10%,16%)" stroke-width="1" stroke-opacity="0.7"'
20svg_mutation_line_style = 'stroke-width="1"'
21svg_crossover_line_style = 'stroke-width="1"'
22svg_spine_line_style = 'stroke="rgb(0%,90%,40%)" stroke-width="2" stroke-opacity="1"'
23svg_scale_line_style = 'stroke="black" stroke-width="0.5" stroke-opacity="1" stroke-dasharray="5, 5"'
24
25svg_dot_style = 'r="2" stroke="black" stroke-width="0.2" fill="red"'
26svg_clear_dot_style = 'r="2" stroke="black" stroke-width="0.4" fill="none"'
27svg_spine_dot_style = 'r="1" stroke="black" stroke-width="0.2" fill="rgb(50%,50%,100%)"'
28
29svg_scale_text_style = 'style="font-family: Arial; font-size: 12; fill: #000000;"'
30
31def hex_to_style(hex):
32    default_style = ' stroke="black" stroke-opacity="0.5" '
33
34    if hex[0] == "#":
35        hex = hex[1:]
36
37    if len(hex) == 6 or len(hex) == 8:
38        try:
39            int(hex, 16)
40        except:
41            print("Invalid characters in the color's hex #" + hex + "! Assuming black.")
42            return default_style
43        red = 100*int(hex[0:2], 16)/255
44        green = 100*int(hex[2:4], 16)/255
45        blue = 100*int(hex[4:6], 16)/255
46        opacity = 0.5
47        if len(hex) == 8:
48            opacity = int(hex[6:8], 16)/255
49        return ' stroke="rgb(' +str(red)+ '%,' +str(green)+ '%,' +str(blue)+ '%)" stroke-opacity="' +str(opacity)+ '" '
50    else:
51        print("Invalid number of digits in the color's hex #" + hex + "! Assuming black.")
52        return default_style
53
54def svg_add_line(from_pos, to_pos, style=svg_line_style):
55    svg_file.write('<line ' + style + ' x1="' + str(from_pos[0]) + '" x2="' + str(to_pos[0]) +
56                   '" y1="' + str(from_pos[1]) + '" y2="' + str(to_pos[1]) + '"  fill="none"/>')
57
58def svg_add_text(text, pos, anchor, style=svg_scale_text_style):
59    svg_file.write('<text ' + style + ' text-anchor="' + anchor + '" x="' + str(pos[0]) + '" y="' + str(pos[1]) + '" >' + text + '</text>')
60
61def svg_add_dot(pos, style=svg_dot_style):
62    svg_file.write('<circle ' + style + ' cx="' + str(pos[0]) + '" cy="' + str(pos[1]) + '" />')
63
64def svg_generate_line_style(percent):
65    # hotdog
66    from_col = [100, 70, 0]
67    to_col = [60, 0, 0]
68    # lava
69    # from_col = [100, 80, 0]
70    # to_col = [100, 0, 0]
71    # neon
72    # from_col = [30, 200, 255]
73    # to_col = [240, 0, 220]
74
75    from_opa = 0.2
76    to_opa = 1.0
77    from_stroke = 1
78    to_stroke = 3
79
80    opa = from_opa*(1-percent) + to_opa*percent
81    stroke = from_stroke*(1-percent) + to_stroke*percent
82
83    percent = 1 - ((1-percent)**20)
84
85    return 'stroke="rgb(' + str(from_col[0]*(1-percent) + to_col[0]*percent) + '%,' \
86           + str(from_col[1]*(1-percent) + to_col[1]*percent) + '%,' \
87           + str(from_col[2]*(1-percent) + to_col[2]*percent) + '%)" stroke-width="' + str(stroke) + '" stroke-opacity="' + str(opa) + '"'
88
89def svg_generate_dot_style(kind):
90    kinds = ["red", "lawngreen", "royalblue", "magenta", "yellow", "cyan", "white", "black"]
91
92    r = min(2500/len(nodes), 10)
93
94    return 'fill="' + kinds[kind] + '" r="' + str(r) + '" stroke="black" stroke-width="' + str(r/10) + '" fill-opacity="1.0" ' \
95           'stroke-opacity="1.0"'
96
97# -------------------
98
99def load_data(dir):
100    global firstnode, nodes, inv_nodes, time
101    f = open(dir)
102    for line in f:
103        sline = line.split(' ', 1)
104        if len(sline) == 2:
105            if sline[0] == "[OFFSPRING]":
106                creature = json.loads(sline[1])
107                #print("B" +str(creature))
108                if "FromIDs" in creature:
109                    if not creature["ID"] in nodes:
110                        nodes[creature["ID"]] = {}
111                        # we assign to each parent its contribution to the genotype of the child
112                        for i in range(0, len(creature["FromIDs"])):
113                            inherited = 1 #(creature["Inherited"][i] if 'Inherited' in creature else 1) #ONLY FOR NOW
114                            nodes[creature["ID"]][creature["FromIDs"][i]] = inherited
115                    else:
116                        print("Duplicated entry for " + creature["ID"])
117                        quit()
118
119                    if not creature["FromIDs"][0] in nodes:
120                        firstnode = creature["FromIDs"][0]
121
122                if "Time" in creature:
123                    time[creature["ID"]] = creature["Time"]
124
125                if "Kind" in creature:
126                    kind[creature["ID"]] = creature["Kind"]
127
128    for k, v in sorted(nodes.items()):
129        for val in sorted(v):
130            inv_nodes[val] = inv_nodes.get(val, [])
131            inv_nodes[val].append(k)
132
133
134def load_simple_data(dir):
135    global firstnode, nodes, inv_nodes
136    f = open(dir)
137    for line in f:
138        sline = line.split()
139        if len(sline) > 1:
140            #if int(sline[0]) > 15000:
141            #    break
142            if sline[0] == firstnode:
143                continue
144            nodes[sline[0]] = str(max(int(sline[1]), int(firstnode)))
145        else:
146            firstnode = sline[0]
147
148    for k, v in sorted(nodes.items()):
149        inv_nodes[v] = inv_nodes.get(v, [])
150        inv_nodes[v].append(k)
151
152    #print(str(inv_nodes))
153    #quit()
154
155def compute_depth(node):
156    my_depth = 0
157    if node in inv_nodes:
158        for c in inv_nodes[node]:
159            my_depth = max(my_depth, compute_depth(c)+1)
160    depth[node] = my_depth
161    return my_depth
162
163# ------------------------------------
164
165def xmin_crowd(x1, x2, y):
166    if BALANCE == "RANDOM":
167        return (x1 if random.randrange(2) == 0 else x2)
168    elif BALANCE == "MIN":
169        x1_closest = 999999
170        x2_closest = 999999
171        for pos in positions:
172            pos = positions[pos]
173            if pos[1] == y:
174                x1_closest = min(x1_closest, abs(x1-pos[0]))
175                x2_closest = min(x2_closest, abs(x2-pos[0]))
176        return (x1 if x1_closest > x2_closest else x2)
177    elif BALANCE == "DENSITY":
178        x1_dist = 0
179        x2_dist = 0
180        for pos in positions:
181            pos = positions[pos]
182            if pos[1] > y-10 or pos[1] < y+10:
183                dy = pos[1]-y
184                dx1 = pos[0]-x1
185                dx2 = pos[0]-x2
186
187                x1_dist += math.sqrt(dy**2 + dx1**2)
188                x2_dist += math.sqrt(dy**2 + dx2**2)
189        return (x1 if x1_dist > x2_dist else x2)
190
191# ------------------------------------
192
193def prepos_children_reccurent(node):
194    global visited
195    for c in inv_nodes[node]:
196
197        # we want to visit the node just once, after all of its parents
198        if not all_parents_visited(c):
199            continue
200        else:
201            visited[c] = True
202
203        cy = 0
204        if TIME == "BIRTHS":
205            if c[0] == "c":
206                cy = int(c[1:])
207            else:
208                cy = int(c)
209        elif TIME == "GENERATIONAL":
210            cy = positions[node][1]+1
211        elif TIME == "REAL":
212            cy = time[c]
213
214        if len(nodes[c]) == 1:
215            dissimilarity = 0
216            if JITTER == True:
217                dissimilarity = random.gauss(0,1)
218            else:
219                dissimilarity = 1
220            positions[c] = [xmin_crowd(positions[node][0]-dissimilarity, positions[node][0]+dissimilarity, cy), cy]
221        else:
222            vsum = sum([v for k, v in nodes[c].items()])
223            cx = sum([positions[k][0]*v/vsum for k, v in nodes[c].items()])
224
225            if JITTER == True:
226                positions[c] = [cx + random.gauss(0, 0.1), cy]
227            else:
228                positions[c] = [cx, cy]
229
230
231        if c in inv_nodes:
232            prepos_children_reccurent(c)
233
234def prepos_children():
235    global max_height, max_width, min_width, visited
236
237    if not bool(time):
238        print("REAL time requested, but no real time data provided. Assuming BIRTHS time instead.")
239        TIME = "BIRTHS"
240
241    positions[firstnode] = [0, 0]
242
243    visited = {}
244    visited[firstnode] = True
245    prepos_children_reccurent(firstnode)
246
247    for pos in positions:
248        max_height = max(max_height, positions[pos][1])
249        max_width = max(max_width, positions[pos][0])
250        min_width = min(min_width, positions[pos][0])
251
252# ------------------------------------
253
254def all_parents_visited(node):
255    apv = True
256    for k, v in sorted(nodes[node].items()):
257        if not k in visited:
258            apv = False
259            break
260    return apv
261# ------------------------------------
262
263def draw_children():
264    max_depth = 0
265    for k, v in depth.items():
266            max_depth = max(max_depth, v)
267
268    nodes_to_visit = [firstnode]
269    while True:
270        current_node = nodes_to_visit[0]
271
272        if current_node in inv_nodes:
273            for c in inv_nodes[current_node]: # inv_node => p->c
274
275                if not c in nodes_to_visit:
276                    nodes_to_visit.append(c)
277
278                line_style = ""
279                if COLORING == "NONE":
280                    line_style = svg_line_style
281                elif COLORING == "TYPE":
282                    line_style = (svg_mutation_line_style if len(nodes[c]) == 1 else svg_crossover_line_style)
283                else: # IMPORTANCE, default
284                    line_style = svg_generate_line_style(depth[c]/max_depth)
285
286                svg_add_line( (w_margin+w_no_margs*(positions[current_node][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[current_node][1]/max_height),
287                        (w_margin+w_no_margs*(positions[c][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[c][1]/max_height), line_style)
288
289        # we want to draw the node just once
290        if DOT_STYLE == "NONE":
291            continue
292        elif DOT_STYLE == "TYPE":
293            dot_style = svg_generate_dot_style(kind[current_node] if current_node in kind else 0) #type
294        else: # NORMAL, default
295            dot_style = svg_clear_dot_style #svg_generate_dot_style(depth[c]/max_depth)
296        svg_add_dot( (w_margin+w_no_margs*(positions[current_node][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[current_node][1]/max_height), dot_style)
297        #svg_add_text( str(depth[current_node]), (w_margin+w_no_margs*(positions[current_node][0]-min_width)/(max_width-min_width),
298        # h_margin+h_no_margs*positions[current_node][1]/max_height), "end")
299
300        # we remove the current node from the list
301        nodes_to_visit = nodes_to_visit[1:]
302        # if none left, we can stop
303        if len(nodes_to_visit) == 0:
304            break
305
306def draw_spine():
307    nodes_to_visit = [firstnode]
308    while True:
309        current_node = nodes_to_visit[0]
310
311        if current_node in inv_nodes:
312            for c in inv_nodes[current_node]: # inv_node => p->c
313                if depth[c] == depth[current_node] - 1:
314                    if not c in nodes_to_visit:
315                        nodes_to_visit.append(c)
316                    line_style = svg_spine_line_style
317                    svg_add_line( (w_margin+w_no_margs*(positions[current_node][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[current_node][1]/max_height),
318                        (w_margin+w_no_margs*(positions[c][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[c][1]/max_height), line_style)
319
320        # we remove the current node from the list
321        nodes_to_visit = nodes_to_visit[1:]
322        # if none left, we can stop
323        if len(nodes_to_visit) == 0:
324            break
325
326def draw_skeleton():
327    nodes_to_visit = [firstnode]
328    while True:
329        current_node = nodes_to_visit[0]
330
331        if current_node in inv_nodes:
332            for c in inv_nodes[current_node]: # inv_node => p->c
333                if depth[c] >= min_skeleton_depth:
334                    if not c in nodes_to_visit:
335                        nodes_to_visit.append(c)
336                    line_style = svg_spine_line_style
337                    svg_add_line( (w_margin+w_no_margs*(positions[current_node][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[current_node][1]/max_height),
338                        (w_margin+w_no_margs*(positions[c][0]-min_width)/(max_width-min_width), h_margin+h_no_margs*positions[c][1]/max_height), line_style)
339
340        # we remove the current node from the list
341        nodes_to_visit = nodes_to_visit[1:]
342        # if none left, we can stop
343        if len(nodes_to_visit) == 0:
344            break
345
346# ------------------------------------
347
348def draw_scale(filename ,type):
349
350    svg_add_text("Generated from " + filename.split("\\")[-1], (5, 15), "start")
351
352    svg_add_line( (w*0.7, h_margin), (w, h_margin), svg_scale_line_style)
353    start_text = ""
354    if TIME == "BIRTHS":
355       start_text = "Birth #" + str(min([int(k[1:]) for k, v in nodes.items()]))
356    if TIME == "REAL":
357       start_text = "Time " + str(min([v for k, v in time.items()]))
358    if TIME == "GENERATIONAL":
359       start_text = "Depth " + str(min([v for k, v in depth.items()]))
360    svg_add_text( start_text, (w, h_margin + 15), "end")
361
362    svg_add_line( (w*0.7, h-h_margin), (w, h-h_margin), svg_scale_line_style)
363    end_text = ""
364    if TIME == "BIRTHS":
365       end_text = "Birth #" + str(max([int(k[1:]) for k, v in nodes.items()]))
366    if TIME == "REAL":
367       end_text = "Time " + str(max([v for k, v in time.items()]))
368    if TIME == "GENERATIONAL":
369       end_text = "Depth " + str(max([v for k, v in depth.items()]))
370    svg_add_text( end_text, (w, h-h_margin + 15), "end")
371
372
373##################################################### main #####################################################
374
375args = 0
376
377h = 800
378w = 600
379h_margin = 20
380w_margin = 10
381h_no_margs = h - 2* h_margin
382w_no_margs = w - 2* w_margin
383
384max_height = 0
385max_width = 0
386min_width = 9999999999
387
388min_skeleton_depth = 0
389
390firstnode = ""
391nodes = {}
392inv_nodes = {}
393positions = {}
394visited= {}
395depth = {}
396time = {}
397kind = {}
398
399def main():
400    global svg_file, min_skeleton_depth, args, \
401        TIME, BALANCE, DOT_STYLE, COLORING, JITTER, \
402        svg_mutation_line_style, svg_crossover_line_style
403
404    parser = argparse.ArgumentParser(description='Draws a genealogical tree (generates a SVG file) based on parent-child relationship information from a text file. Supports files generated by Framsticks experiments.')
405    parser.add_argument('-i', '--in', dest='input', required=True, help='input file name with stuctured evolutionary data')
406    parser.add_argument('-o', '--out', dest='output', required=True, help='output file name for the evolutionary tree (SVG format)')
407    draw_tree_parser = parser.add_mutually_exclusive_group(required=False)
408    draw_tree_parser.add_argument('--draw-tree', dest='draw_tree', action='store_true', help='whether drawing the full tree should be skipped')
409    draw_tree_parser.add_argument('--no-draw-tree', dest='draw_tree', action='store_false')
410
411    draw_skeleton_parser = parser.add_mutually_exclusive_group(required=False)
412    draw_skeleton_parser.add_argument('--draw-skeleton', dest='draw_skeleton', action='store_true', help='whether the skeleton of the tree should be drawn')
413    draw_skeleton_parser.add_argument('--no-draw-skeleton', dest='draw_skeleton', action='store_false')
414
415    draw_spine_parser = parser.add_mutually_exclusive_group(required=False)
416    draw_spine_parser.add_argument('--draw-spine', dest='draw_spine', action='store_true', help='whether the spine of the tree should be drawn')
417    draw_spine_parser.add_argument('--no-draw-spine', dest='draw_spine', action='store_false')
418
419    #TODO: better names for those parameters
420    parser.add_argument('-t', '--time', default='GENERATIONAL', dest='time', help='values on vertical axis (BIRTHS/GENERATIONAL/REAL); '
421                                                                      'BIRTHS: time measured as the number of births since the beginning; '
422                                                                      'GENERATIONAL: time measured as number of ancestors; '
423                                                                      'REAL: real time of the simulation')
424    parser.add_argument('-b', '--balance', default='DENSITY', dest='balance', help='method of placing nodes in the tree (RANDOM/MIN/DENSITY)')
425    parser.add_argument('-s', '--scale', default='NONE', dest='scale', help='type of timescale added to the tree (NONE/SIMPLE)')
426    parser.add_argument('-c', '--coloring', default='IMPORTANCE', dest="coloring", help='method of coloring the tree (NONE/IMPORTANCE/TYPE)')
427    parser.add_argument('-d', '--dots', default='TYPE', dest='dots', help='method of drawing dots (individuals) (NONE/NORMAL/TYPE)')
428    parser.add_argument('-j', '--jitter', dest="jitter", action='store_true', help='draw horizontal positions of children from the normal distribution')
429
430    parser.add_argument('--color-mut', default="#000000", dest="color_mut", help='color of clone/mutation lines in rgba (e.g. #FF60B240) for TYPE coloring')
431    parser.add_argument('--color-cross', default="#660198", dest="color_cross", help='color of crossover lines in rgba (e.g. #FF60B240) for TYPE coloring')
432
433    parser.add_argument('--min-skeleton-depth', type=int, default=2, dest='min_skeleton_depth', help='minimal distance from the leafs for the nodes in the skeleton')
434    parser.add_argument('--seed', type=int, dest='seed', help='seed for the random number generator (-1 for random)')
435
436    parser.add_argument('--simple-data', type=bool, dest='simple_data', help='input data are given in a simple format (#child #parent)')
437
438    parser.set_defaults(draw_tree=True)
439    parser.set_defaults(draw_skeleton=False)
440    parser.set_defaults(draw_spine=False)
441
442    parser.set_defaults(seed=-1)
443
444    args = parser.parse_args()
445
446    TIME = args.time
447    BALANCE = args.balance
448    DOT_STYLE = args.dots
449    COLORING = args.coloring
450    JITTER = args.jitter
451
452    svg_mutation_line_style += hex_to_style(args.color_mut)
453    svg_crossover_line_style += hex_to_style(args.color_cross)
454
455    dir = args.input
456    min_skeleton_depth = args.min_skeleton_depth
457    seed = args.seed
458    if seed == -1:
459        seed = random.randint(0, 10000)
460    random.seed(seed)
461    print("seed:", seed)
462
463    if args.simple_data:
464        load_simple_data(dir)
465    else:
466        load_data(dir)
467
468    compute_depth(firstnode)
469
470    svg_file = open(args.output, "w")
471    svg_file.write('<svg xmlns:svg="http://www.w3.org/2000/svg" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" version="1.0" '
472                   'width="' + str(w) + '" height="' + str(h) + '">')
473
474    prepos_children()
475
476    if args.draw_tree:
477        draw_children()
478    if args.draw_skeleton:
479        draw_skeleton()
480    if args.draw_spine:
481        draw_spine()
482
483    draw_scale(dir, args.scale)
484
485    svg_file.write("</svg>")
486    svg_file.close()
487
488main()
Note: See TracBrowser for help on using the repository browser.