source: mds-and-trees/tree-genealogy.py @ 621

# Draws a genealogical tree (generates a SVG file) based on parent-child relationship information.
# Supports files generated by Framsticks experiments.

import json
import random
import math
import argparse
import time as ttime

TIME = "" # BIRTHS / GENERATIONAL / REAL
BALANCE = "" # MIN / DENSITY

DOT_STYLE = "" # NONE / NORMAL / CLEAR

JITTER = "" #

# ------SVG---------
svg_file = 0

svg_line_style = 'stroke="rgb(90%,10%,16%)" stroke-width="1" stroke-opacity="0.7"'
svg_mutation_line_style = 'stroke-width="1"'
svg_crossover_line_style = 'stroke-width="1"'
svg_spine_line_style = 'stroke="rgb(0%,90%,40%)" stroke-width="2" stroke-opacity="1"'
svg_scale_line_style = 'stroke="black" stroke-width="0.5" stroke-opacity="1" stroke-dasharray="5, 5"'

svg_dot_style = 'r="2" stroke="black" stroke-width="0.2" fill="red"'
svg_clear_dot_style = 'r="2" stroke="black" stroke-width="0.4" fill="none"'
svg_spine_dot_style = 'r="1" stroke="black" stroke-width="0.2" fill="rgb(50%,50%,100%)"'

svg_scale_text_style = 'style="font-family: Arial; font-size: 12; fill: #000000;"'

def hex_to_style(hex):
    default_style = ' stroke="black" stroke-opacity="0.5" '

    if hex[0] == "#":
        hex = hex[1:]

    if len(hex) == 6 or len(hex) == 8:
        try:
            int(hex, 16)
        except:
            print("Invalid characters in the color's hex #" + hex + "! Assuming black.")
            return default_style
        red = 100*int(hex[0:2], 16)/255
        green = 100*int(hex[2:4], 16)/255
        blue = 100*int(hex[4:6], 16)/255
        opacity = 0.5
        if len(hex) == 8:
            opacity = int(hex[6:8], 16)/255
        return ' stroke="rgb(' +str(red)+ '%,' +str(green)+ '%,' +str(blue)+ '%)" stroke-opacity="' +str(opacity)+ '" '
    else:
        print("Invalid number of digits in the color's hex #" + hex + "! Assuming black.")
        return default_style

def svg_add_line(from_pos, to_pos, style=svg_line_style):
    svg_file.write('<line ' + style + ' x1="' + str(from_pos[0]) + '" x2="' + str(to_pos[0]) +
                   '" y1="' + str(from_pos[1]) + '" y2="' + str(to_pos[1]) + '"  fill="none"/>')

def svg_add_text(text, pos, anchor, style=svg_scale_text_style):
    svg_file.write('<text ' + style + ' text-anchor="' + anchor + '" x="' + str(pos[0]) + '" y="' + str(pos[1]) + '" >' + text + '</text>')

def svg_add_dot(pos, style=svg_dot_style):
    svg_file.write('<circle ' + style + ' cx="' + str(pos[0]) + '" cy="' + str(pos[1]) + '" />')

def svg_generate_line_style(percent):
    # hotdog
    from_col = [100, 70, 0]
    to_col = [60, 0, 0]
    # lava
    # from_col = [100, 80, 0]
    # to_col = [100, 0, 0]
    # neon
    # from_col = [30, 200, 255]
    # to_col = [240, 0, 220]

    from_opa = 0.2
    to_opa = 1.0
    from_stroke = 1
    to_stroke = 3

    opa = from_opa*(1-percent) + to_opa*percent
    stroke = from_stroke*(1-percent) + to_stroke*percent

    percent = 1 - ((1-percent)**20)

    return 'stroke="rgb(' + str(from_col[0]*(1-percent) + to_col[0]*percent) + '%,' \
           + str(from_col[1]*(1-percent) + to_col[1]*percent) + '%,' \
           + str(from_col[2]*(1-percent) + to_col[2]*percent) + '%)" stroke-width="' + str(stroke) + '" stroke-opacity="' + str(opa) + '"'

def svg_generate_dot_style(kind):
    kinds = ["red", "lawngreen", "royalblue", "magenta", "yellow", "cyan", "white", "black"]

    r = min(2500/len(nodes), 10)

    return 'fill="' + kinds[kind] + '" r="' + str(r) + '" stroke="black" stroke-width="' + str(r/10) + '" fill-opacity="1.0" ' \
           'stroke-opacity="1.0"'

# -------------------

def load_data(dir):
    global firstnode, nodes, inv_nodes, time
    f = open(dir)
    for line in f:
        sline = line.split(' ', 1)
        if len(sline) == 2:
            if sline[0] == "[OFFSPRING]":
                creature = json.loads(sline[1])
                #print("B" +str(creature))
                if "FromIDs" in creature:
                    if not creature["ID"] in nodes:
                        nodes[creature["ID"]] = {}
                        # we assign to each parent its contribution to the genotype of the child
                        for i in range(0, len(creature["FromIDs"])):
                            inherited = 1 #(creature["Inherited"][i] if 'Inherited' in creature else 1) #ONLY FOR NOW
                            nodes[creature["ID"]][creature["FromIDs"][i]] = inherited
                    else:
                        print("Duplicated entry for " + creature["ID"])
                        quit()

                    if not creature["FromIDs"][0] in nodes and firstnode == None:
                        firstnode = creature["FromIDs"][0]

                if "Time" in creature:
                    time[creature["ID"]] = creature["Time"]

                if "Kind" in creature:
                    kind[creature["ID"]] = creature["Kind"]

    for k, v in sorted(nodes.items()):
        for val in sorted(v):
            inv_nodes[val] = inv_nodes.get(val, [])
            inv_nodes[val].append(k)

    print(len(nodes))


def load_simple_data(dir):
    global firstnode, nodes, inv_nodes
    f = open(dir)
    for line in f:
        sline = line.split()
        if len(sline) > 1:
            #if int(sline[0]) > 15000:
            #    break
            if sline[0] == firstnode:
                continue
            nodes[sline[0]] = str(max(int(sline[1]), int(firstnode)))
        else:
            firstnode = sline[0]

    for k, v in sorted(nodes.items()):
        inv_nodes[v] = inv_nodes.get(v, [])
        inv_nodes[v].append(k)

    #print(str(inv_nodes))
    #quit()

def compute_depth(node):
    my_depth = 0
    if node in inv_nodes:
        for c in inv_nodes[node]:
            my_depth = max(my_depth, compute_depth(c)+1)
    depth[node] = my_depth
    return my_depth

# ------------------------------------

def xmin_crowd(x1, x2, y):
    if BALANCE == "RANDOM":
        return (x1 if random.randrange(2) == 0 else x2)
    elif BALANCE == "MIN":
        x1_closest = 999999
        x2_closest = 999999
        for pos in positions:
            pos = positions[pos]
            if pos[1] == y:
                x1_closest = min(x1_closest, abs(x1-pos[0]))
                x2_closest = min(x2_closest, abs(x2-pos[0]))
        return (x1 if x1_closest > x2_closest else x2)
    elif BALANCE == "DENSITY":
        x1_dist = 0
        x2_dist = 0
        for pos in positions:
            pos = positions[pos]
            if pos[1] > y-10 or pos[1] < y+10:
                dy = pos[1]-y
                dx1 = pos[0]-x1
                dx2 = pos[0]-x2

                x1_dist += math.sqrt(dy**2 + dx1**2)
                x2_dist += math.sqrt(dy**2 + dx2**2)
        return (x1 if x1_dist > x2_dist else x2)

# ------------------------------------

# def prepos_children_reccurent(node):
#     global visited
#     for c in inv_nodes[node]:
#
#         # we want to visit the node just once, after all of its parents
#         if not all_parents_visited(c):
#             continue
#         else:
#             visited[c] = True
#
#         cy = 0
#         if TIME == "BIRTHS":
#             if c[0] == "c":
#                 cy = int(c[1:])
#             else:
#                 cy = int(c)
#         elif TIME == "GENERATIONAL":
#             cy = positions[node][1]+1
#         elif TIME == "REAL":
#             cy = time[c]
#
#         if len(nodes[c]) == 1:
#             dissimilarity = 0
#             if JITTER == True:
#                 dissimilarity = random.gauss(0,1)
#             else:
#                 dissimilarity = 1
#             positions[c] = [xmin_crowd(positions[node][0]-dissimilarity, positions[node][0]+dissimilarity, cy), cy]
#         else:
#             vsum = sum([v for k, v in nodes[c].items()])
#             cx = sum([positions[k][0]*v/vsum for k, v in nodes[c].items()])
#
#             if JITTER == True:
#                 positions[c] = [cx + random.gauss(0, 0.1), cy]
#             else:
#                 positions[c] = [cx, cy]
#
#
#         if c in inv_nodes:
#             prepos_children_reccurent(c)

def prepos_children():
    global max_height, max_width, min_width, visited, TIME

    print("firstnode " + firstnode)

    if not bool(time):
        print("REAL time requested, but no real time data provided. Assuming BIRTHS time instead.")
        TIME = "BIRTHS"

    positions[firstnode] = [0, 0]

    #visited = {}
    #visited[firstnode] = True


    nodes_to_visit = [firstnode]

    ccc = 0
    timet = ttime.time()

    while True:

        ccc += 1
        if ccc%1000 == 0 :
            print(str(ccc) + " "  + str(ttime.time()-timet))
            timet = ttime.time()

        current_node = nodes_to_visit[0]

        if current_node in inv_nodes:
            for c in inv_nodes[current_node]:
                # we want to visit the node just once, after all of its parents
                if c not in nodes_to_visit:
                    nodes_to_visit.append(c)

                    cy = 0
                    if TIME == "BIRTHS":
                        if c[0] == "c":
                            cy = int(c[1:])
                        else:
                            cy = int(c)
                    elif TIME == "GENERATIONAL":
                        cy = positions[current_node][1]+1
                    elif TIME == "REAL":
                        cy = time[c]

                    if len(nodes[c]) == 1:
                        dissimilarity = 0
                        if JITTER == True:
                            dissimilarity = random.gauss(0,1)
                        else:
                            dissimilarity = 1
                        positions[c] = [xmin_crowd(positions[current_node][0]-dissimilarity, positions[current_node][0]+dissimilarity, cy), cy]
                    else:
                        vsum = sum([v for k, v in nodes[c].items()])
                        cx = sum([positions[k][0]*v/vsum for k, v in nodes[c].items()])

                        if JITTER == True:
                            positions[c] = [cx + random.gauss(0, 0.1), cy]
                        else:
                            positions[c] = [cx, cy]


        #if c in inv_nodes:
        #    prepos_children_reccurent(c)

        nodes_to_visit = nodes_to_visit[1:]
        # if none left, we can stop
        if len(nodes_to_visit) == 0:
            break


   # prepos_children_reccurent(firstnode)

    for pos in positions:
        max_height = max(max_height, positions[pos][1])
        max_width = max(max_width, positions[pos][0])
        min_width = min(min_width, positions[pos][0])

# ------------------------------------

def all_parents_visited(node):
    apv = True
    for k, v in sorted(nodes[node].items()):
        if not k in visited:
            apv = False
            break
    return apv
# ------------------------------------

def draw_children():
    max_depth = 0
    for k, v in depth.items():
            max_depth = max(max_depth, v)

    nodes_to_visit = [firstnode]
    while True:
        current_node = nodes_to_visit[0]

        if current_node in inv_nodes:
            for c in inv_nodes[current_node]: # inv_node => p->c

                if not c in nodes_to_visit:
                    nodes_to_visit.append(c)

                line_style = ""
                if COLORING == "NONE":
                    line_style = svg_line_style
                elif COLORING == "TYPE":
                    line_style = (svg_mutation_line_style if len(nodes[c]) == 1 else svg_crossover_line_style)
                else: # IMPORTANCE, default
                    line_style = svg_generate_line_style(depth[c]/max_depth)

                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),
                        (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)

        # we want to draw the node just once
        if DOT_STYLE == "NONE":
            continue
        elif DOT_STYLE == "TYPE":
            dot_style = svg_generate_dot_style(kind[current_node] if current_node in kind else 0) #type
        else: # NORMAL, default
            dot_style = svg_clear_dot_style #svg_generate_dot_style(depth[c]/max_depth)
        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)
        #svg_add_text( str(depth[current_node]), (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), "end")

        # we remove the current node from the list
        nodes_to_visit = nodes_to_visit[1:]
        # if none left, we can stop
        if len(nodes_to_visit) == 0:
            break

def draw_spine():
    nodes_to_visit = [firstnode]
    while True:
        current_node = nodes_to_visit[0]

        if current_node in inv_nodes:
            for c in inv_nodes[current_node]: # inv_node => p->c
                if depth[c] == depth[current_node] - 1:
                    if not c in nodes_to_visit:
                        nodes_to_visit.append(c)
                    line_style = svg_spine_line_style
                    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),
                        (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)

        # we remove the current node from the list
        nodes_to_visit = nodes_to_visit[1:]
        # if none left, we can stop
        if len(nodes_to_visit) == 0:
            break

def draw_skeleton():
    nodes_to_visit = [firstnode]
    while True:
        current_node = nodes_to_visit[0]

        if current_node in inv_nodes:
            for c in inv_nodes[current_node]: # inv_node => p->c
                if depth[c] >= min_skeleton_depth:
                    if not c in nodes_to_visit:
                        nodes_to_visit.append(c)
                    line_style = svg_spine_line_style
                    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),
                        (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)

        # we remove the current node from the list
        nodes_to_visit = nodes_to_visit[1:]
        # if none left, we can stop
        if len(nodes_to_visit) == 0:
            break

# ------------------------------------

def draw_scale(filename ,type):

    svg_add_text("Generated from " + filename.split("\\")[-1], (5, 15), "start")

    svg_add_line( (w*0.7, h_margin), (w, h_margin), svg_scale_line_style)
    start_text = ""
    if TIME == "BIRTHS":
       start_text = "Birth #" + str(min([int(k[1:]) for k, v in nodes.items()]))
    if TIME == "REAL":
       start_text = "Time " + str(min([v for k, v in time.items()]))
    if TIME == "GENERATIONAL":
       start_text = "Depth " + str(min([v for k, v in depth.items()]))
    svg_add_text( start_text, (w, h_margin + 15), "end")

    svg_add_line( (w*0.7, h-h_margin), (w, h-h_margin), svg_scale_line_style)
    end_text = ""
    if TIME == "BIRTHS":
       end_text = "Birth #" + str(max([int(k[1:]) for k, v in nodes.items()]))
    if TIME == "REAL":
       end_text = "Time " + str(max([v for k, v in time.items()]))
    if TIME == "GENERATIONAL":
       end_text = "Depth " + str(max([v for k, v in depth.items()]))
    svg_add_text( end_text, (w, h-h_margin + 15), "end")


##################################################### main #####################################################

args = 0

h = 800
w = 600
h_margin = 20
w_margin = 10
h_no_margs = h - 2* h_margin
w_no_margs = w - 2* w_margin

max_height = 0
max_width = 0
min_width = 9999999999

min_skeleton_depth = 0

firstnode = None
nodes = {}
inv_nodes = {}
positions = {}
visited= {}
depth = {}
time = {}
kind = {}

def main():
    global svg_file, min_skeleton_depth, args, \
        TIME, BALANCE, DOT_STYLE, COLORING, JITTER, \
        svg_mutation_line_style, svg_crossover_line_style

    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.')
    parser.add_argument('-i', '--in', dest='input', required=True, help='input file name with stuctured evolutionary data')
    parser.add_argument('-o', '--out', dest='output', required=True, help='output file name for the evolutionary tree (SVG format)')
    draw_tree_parser = parser.add_mutually_exclusive_group(required=False)
    draw_tree_parser.add_argument('--draw-tree', dest='draw_tree', action='store_true', help='whether drawing the full tree should be skipped')
    draw_tree_parser.add_argument('--no-draw-tree', dest='draw_tree', action='store_false')

    draw_skeleton_parser = parser.add_mutually_exclusive_group(required=False)
    draw_skeleton_parser.add_argument('--draw-skeleton', dest='draw_skeleton', action='store_true', help='whether the skeleton of the tree should be drawn')
    draw_skeleton_parser.add_argument('--no-draw-skeleton', dest='draw_skeleton', action='store_false')

    draw_spine_parser = parser.add_mutually_exclusive_group(required=False)
    draw_spine_parser.add_argument('--draw-spine', dest='draw_spine', action='store_true', help='whether the spine of the tree should be drawn')
    draw_spine_parser.add_argument('--no-draw-spine', dest='draw_spine', action='store_false')

    #TODO: better names for those parameters
    parser.add_argument('-t', '--time', default='GENERATIONAL', dest='time', help='values on vertical axis (BIRTHS/GENERATIONAL(d)/REAL); '
                                                                      'BIRTHS: time measured as the number of births since the beginning; '
                                                                      'GENERATIONAL: time measured as number of ancestors; '
                                                                      'REAL: real time of the simulation')
    parser.add_argument('-b', '--balance', default='DENSITY', dest='balance', help='method of placing nodes in the tree (RANDOM/MIN/DENSITY(d))')
    parser.add_argument('-s', '--scale', default='NONE', dest='scale', help='type of timescale added to the tree (NONE(d)/SIMPLE)')
    parser.add_argument('-c', '--coloring', default='IMPORTANCE', dest="coloring", help='method of coloring the tree (NONE/IMPORTANCE(d)/TYPE)')
    parser.add_argument('-d', '--dots', default='TYPE', dest='dots', help='method of drawing dots (individuals) (NONE/NORMAL/TYPE(d))')
    parser.add_argument('-j', '--jitter', dest="jitter", action='store_true', help='draw horizontal positions of children from the normal distribution')

    parser.add_argument('--color-mut', default="#000000", dest="color_mut", help='color of clone/mutation lines in rgba (e.g. #FF60B240) for TYPE coloring')
    parser.add_argument('--color-cross', default="#660198", dest="color_cross", help='color of crossover lines in rgba (e.g. #FF60B240) for TYPE coloring')

    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')
    parser.add_argument('--seed', type=int, dest='seed', help='seed for the random number generator (-1 for random)')

    parser.add_argument('--simple-data', type=bool, dest='simple_data', help='input data are given in a simple format (#child #parent)')

    parser.set_defaults(draw_tree=True)
    parser.set_defaults(draw_skeleton=False)
    parser.set_defaults(draw_spine=False)

    parser.set_defaults(seed=-1)

    args = parser.parse_args()

    TIME = args.time.upper()
    BALANCE = args.balance.upper()
    DOT_STYLE = args.dots.upper()
    COLORING = args.coloring.upper()
    SCALE = args.scale.upper()
    JITTER = args.jitter
    if not TIME in ['BIRTHS', 'GENERATIONAL', 'REAL']\
        or not BALANCE in ['RANDOM', 'MIN', 'DENSITY']\
        or not DOT_STYLE in ['NONE', 'NORMAL', 'TYPE']\
        or not COLORING in ['NONE', 'IMPORTANCE', 'TYPE']\
        or not SCALE in ['NONE', 'SIMPLE']:
        print("Incorrect value of one of the parameters! Closing the program.") #TODO don't be lazy, figure out which parameter is wrong...
        return


    svg_mutation_line_style += hex_to_style(args.color_mut)
    svg_crossover_line_style += hex_to_style(args.color_cross)

    dir = args.input
    min_skeleton_depth = args.min_skeleton_depth
    seed = args.seed
    if seed == -1:
        seed = random.randint(0, 10000)
    random.seed(seed)
    print("seed:", seed)

    if args.simple_data:
        load_simple_data(dir)
    else:
        load_data(dir)

    compute_depth(firstnode)

    svg_file = open(args.output, "w")
    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" '
                   'width="' + str(w) + '" height="' + str(h) + '">')

    prepos_children()

    if args.draw_tree:
        draw_children()
    if args.draw_skeleton:
        draw_skeleton()
    if args.draw_spine:
        draw_spine()

    draw_scale(dir, SCALE)

    svg_file.write("</svg>")
    svg_file.close()

main()