treesak 1.51.2__py3-none-any.whl

TreeSAK/mcmc2tree.py

@@ -0,0 +1,58 @@
+import os
+import argparse
+from ete3 import Tree
+
+
+mcmc2tree_usage = '''
+============ mcmc2tree example commands ============
+
+TreeSAK mcmc2tree -i mcmctree.out -o renamed.tree
+
+====================================================
+'''
+
+
+def mcmc2tree(args):
+
+    mamctree_out = args['i']
+    tree_file    = args['o']
+
+    if os.path.isfile(mamctree_out) is False:
+        print('%s not found, program exited!' % mamctree_out)
+
+    # get tree string from mcmctree_out
+    tree_str = ''
+    tree_line = 0
+    current_line = 1
+    for each_line in open(mamctree_out):
+        if 'Species tree for FigTree.  Branch lengths = posterior mean times; 95% CIs = labels' in each_line:
+            tree_line = current_line + 1
+        if tree_line == current_line:
+            tree_str = each_line.strip()
+        current_line += 1
+
+    tree_str_no_space = tree_str.replace(' ', '')
+
+    # rename tree nodes
+    t = Tree(tree_str_no_space, format=1)
+    for each_node in t.traverse():
+        if each_node.is_leaf():
+            node_name_new = '_'.join(each_node.name.split('_')[1:])
+        else:
+            node_name_new = 't_n%s' % each_node.name
+        each_node.name = node_name_new
+
+    tree_str_renamed = t.write(format=8)
+
+    tree_file_handle = open(tree_file, 'w')
+    tree_file_handle.write(tree_str_renamed + '\n')
+    tree_file_handle.close()
+
+
+if __name__ == '__main__':
+
+    mcmc2tree_parser = argparse.ArgumentParser()
+    mcmc2tree_parser.add_argument('-i',  required=True,  help='the .out file from mcmctree')
+    mcmc2tree_parser.add_argument('-o',  required=True,  help='output tree file')
+    args = vars(mcmc2tree_parser.parse_args())
+    mcmc2tree(args)

TreeSAK/mcmcTC copy.py

@@ -0,0 +1,92 @@
+import argparse
+from ete3 import Tree
+
+
+mcmcTC_usage = '''
+===================== mcmcTC example commands =====================
+
+TreeSAK mcmcTC -i in.tree -o out.tree -tc time_constraints.txt
+
+# Format of constraint file (tab separated columns)
+IMG2264867070_yang,GCF900696045_1_yang	3.46-4.38	Archaeal root
+GCF000015225_1_yang,GCF000007225_1_yang	-2.32	Oxygen Age Constraint, Thermoproteales
+GCF000213215_1_yang,GCA000024305_1_yang	-2.32	Oxygen Age Constraint, Sulfolobales
+GCF000152265_2_yang,GCF000195915_1_yang	-2.32	Oxygen Age Constraint, Thermoplasma
+GCF000376445_1_yang,GCF000172995_2_yang	-1.579	Chitin Age Constraint, Halobacteriales
+GCF000195935_2_yang,GCF000151105_2_yang	-1.579	Chitin Age Constraint, Thermococcales
+GCA000802205_2_yang,GCA000200715_1	0.75-1.49	HGT from Viridiplantae to Thaumarchaeota
+
+===================================================================
+'''
+
+
+def mcmcTC(args):
+
+    tree_file_in        = args['i']
+    time_constraint_txt = args['tc']
+    tree_file_out       = args['o']
+
+    constraint_set = set()
+    constraint_dict = dict()
+    not_recognizable_time_constraint_set = set()
+    for each_constraint in open(time_constraint_txt):
+        each_constraint_split = each_constraint.strip().split('\t')
+        leaf_ids = each_constraint_split[0]
+        provided_age = each_constraint_split[1]
+
+        str_to_add = ''
+        if provided_age.startswith('-'):
+            str_to_add = '<%s' % provided_age[1:]
+        elif provided_age.endswith('-'):
+            str_to_add = '>%s' % provided_age[:-1]
+        elif '-' in provided_age:
+            provided_age_split = provided_age.split('-')
+            str_to_add = '>%s<%s' % (provided_age_split[0], provided_age_split[1])
+        else:
+            not_recognizable_time_constraint_set.add(provided_age)
+
+        constraint_dict[leaf_ids] = str_to_add
+        constraint_set.add(str_to_add)
+
+    if len(not_recognizable_time_constraint_set) > 0:
+        print('Format of the following time constraints are not recognizable, program exited')
+        print(','.join(not_recognizable_time_constraint_set))
+        exit()
+
+    # read in tree
+    tree_in = Tree(tree_file_in, quoted_node_names=True, format=1)
+
+    # add time constraints as node name
+    for each_node in constraint_dict:
+        node_age = constraint_dict[each_node]
+        node_split = each_node.split(',')
+        current_lca = tree_in.get_common_ancestor(node_split)
+        current_lca.add_features(custom_label=node_age)
+        current_lca.name = node_age
+
+    tree_out_str = tree_in.write(format=1)
+
+    # remove branch length of 1
+    tree_out_str = tree_out_str.replace(':1', '')
+
+    # quote constraint strings
+    for each_constraint in constraint_set:
+        tree_out_str = tree_out_str.replace(each_constraint, ("'%s'" % each_constraint))
+
+    # remove the most outside parenthesis
+    tree_out_str = tree_out_str[1:].replace(');', ';')
+
+    # write tree to file
+    with open(tree_file_out, 'w') as tree_file_out_handle:
+        tree_file_out_handle.write('%s\t1\n' % len(tree_in.get_leaves()))
+        tree_file_out_handle.write(tree_out_str)
+
+
+if __name__ == '__main__':
+
+    mcmcTC_parser = argparse.ArgumentParser()
+    mcmcTC_parser.add_argument('-i',   required=True,  help='input tree')
+    mcmcTC_parser.add_argument('-o',   required=True,  help='output tree')
+    mcmcTC_parser.add_argument('-tc',  required=True,  help='time constraint file')
+    args = vars(mcmcTC_parser.parse_args())
+    mcmcTC(args)

TreeSAK/mcmcTC.py

@@ -0,0 +1,104 @@
+import argparse
+from ete3 import Tree
+
+
+mcmcTC_usage = '''
+======================================= mcmcTC example commands =======================================
+
+TreeSAK mcmcTC -i in.tree -o out.tree -tc time_constraints.txt
+
+# Format of constraint file (tab separated columns)
+IMG2264867070_yang,GCF900696045_1_yang	lca	3.46-4.38	Archaeal root
+GCF000015225_1_yang,GCF000007225_1_yang	lca	-2.32	Oxygen Age Constraint, Thermoproteales
+GCF000213215_1_yang,GCA000024305_1_yang	lca	-2.32	Oxygen Age Constraint, Sulfolobales
+GCF000152265_2_yang,GCF000195915_1_yang	lca	-2.32	Oxygen Age Constraint, Thermoplasma
+GCF000376445_1_yang,GCF000172995_2_yang	lca	-1.579	Chitin Age Constraint, Halobacteriales
+GCF000195935_2_yang,GCF000151105_2_yang	lca	-1.579	Chitin Age Constraint, Thermococcales
+GCA000802205_2_yang,GCF000303155_1_yang	lca_p	0.75-1.49	HGT from Viridiplantae to Thaumarchaeota
+
+# lca: last common ancestor
+# lca_p: parent of last common ancestor
+
+=======================================================================================================
+'''
+
+
+def mcmcTC(args):
+
+    tree_file_in        = args['i']
+    time_constraint_txt = args['tc']
+    tree_file_out       = args['o']
+
+    constraint_set = set()
+    constraint_dict = dict()
+    constraint_type_dict = dict()
+    not_recognizable_time_constraint_set = set()
+    for each_constraint in open(time_constraint_txt):
+        each_constraint_split = each_constraint.strip().split('\t')
+        leaf_ids        = each_constraint_split[0]
+        constraint_type = each_constraint_split[1]
+        provided_age    = each_constraint_split[2]
+
+        str_to_add = ''
+        if provided_age.startswith('-'):
+            str_to_add = '<%s' % provided_age[1:]
+        elif provided_age.endswith('-'):
+            str_to_add = '>%s' % provided_age[:-1]
+        elif '-' in provided_age:
+            provided_age_split = provided_age.split('-')
+            str_to_add = '>%s<%s' % (provided_age_split[0], provided_age_split[1])
+        else:
+            not_recognizable_time_constraint_set.add(provided_age)
+
+        constraint_set.add(str_to_add)
+        constraint_dict[leaf_ids] = str_to_add
+        constraint_type_dict[leaf_ids] = constraint_type
+
+    if len(not_recognizable_time_constraint_set) > 0:
+        print('Format of the following time constraints are not recognizable, program exited')
+        print(','.join(not_recognizable_time_constraint_set))
+        exit()
+
+    # read in tree
+    tree_in = Tree(tree_file_in, quoted_node_names=True, format=1)
+
+    # add time constraints as node name
+    for each_node in constraint_dict:
+        node_age = constraint_dict[each_node]
+        node_split = each_node.split(',')
+        current_lca = tree_in.get_common_ancestor(node_split)
+        constraint_type = constraint_type_dict[each_node]
+        if constraint_type == 'lca':
+            current_lca.add_features(custom_label=node_age)
+            current_lca.name = node_age
+        elif constraint_type == 'lca_p':
+            current_lca_p = current_lca.up
+            current_lca_p.add_features(custom_label=node_age)
+            current_lca_p.name = node_age
+
+    tree_out_str = tree_in.write(format=1)
+
+    # remove branch length of 1
+    tree_out_str = tree_out_str.replace(':1', '')
+
+    # quote constraint strings
+    for each_constraint in constraint_set:
+        tree_out_str = tree_out_str.replace(each_constraint, ("'%s'" % each_constraint))
+
+    # remove the most outside parenthesis
+    tree_out_str = tree_out_str[1:].replace(');', ';')
+
+    # write tree to file
+    with open(tree_file_out, 'w') as tree_file_out_handle:
+        tree_file_out_handle.write('%s\t1\n' % len(tree_in.get_leaves()))
+        tree_file_out_handle.write(tree_out_str)
+
+
+if __name__ == '__main__':
+
+    mcmcTC_parser = argparse.ArgumentParser()
+    mcmcTC_parser.add_argument('-i',   required=True,  help='input tree')
+    mcmcTC_parser.add_argument('-o',   required=True,  help='output tree')
+    mcmcTC_parser.add_argument('-tc',  required=True,  help='time constraint file')
+    args = vars(mcmcTC_parser.parse_args())
+    mcmcTC(args)

TreeSAK/mcmctree_vs_reltime.R

@@ -0,0 +1,44 @@
+library(ggplot2)
+library(optparse)
+
+
+plot_grouped_HPD95 <- function(data_file, plot_width, plot_height, plot_file){
+
+  dat <- read.table(data_file, header = T)
+
+  ggplot(dat, aes(x = Node, y = Mean, ymin = Low, ymax = High)) +
+    geom_pointrange(aes(col = factor(ColorBy)),
+                    position=position_dodge(width=0.6),                         # controls distance between groups
+                    linewidth = 0.9,                                            # line width
+                    size=0.75) +                                                # size of shape
+    theme_bw() +                                                                # remove background
+    theme(panel.grid.major=element_blank(),                                     # remove grid
+          panel.grid.minor=element_blank()) +                                   # remove grid
+    xlab("") +                                                                  # x-axis label text
+    ylab("95% HPD CI") +                                                        # y-axis label text
+    theme(axis.text.x=element_text(size=12, color='black', angle=30, hjust=1),  # x-axis label, rotate at an angle of 45
+          axis.text.y=element_text(size=12, color='black'),                     # y-axis label
+          legend.text=element_text(size=10)) +                                  # legend label
+    scale_color_discrete(name="Color") +                                        # customize color legend, title
+    guides(color=guide_legend(override.aes=list(linetype=0))) +                   # customize color legend
+    coord_flip() # # plot vertically
+  
+  # write to file
+  ggsave(plot_file, width=plot_width, height=plot_height, dpi=300)
+}
+
+
+option_list = list(
+  make_option(c("-i", "--datain"),  type="character", default=NULL, help="input data matrix"),
+  make_option(c("-x", "--width"),   type="double",    default=8,    help="plot width"),
+  make_option(c("-y", "--height"),  type="double",    default=5,    help="plot height"),
+  make_option(c("-o", "--plotout"), type="character", default=NULL, help="output plot"));
+
+opt_parser      = OptionParser(option_list=option_list);
+opt             = parse_args(opt_parser);
+data_matrix_txt = opt$datain
+plot_width      = opt$width
+plot_height     = opt$height
+output_plot     = opt$plotout
+
+plot_grouped_HPD95(data_matrix_txt, plot_width, plot_height, output_plot)

TreeSAK/mcmctree_vs_reltime.py

@@ -0,0 +1,252 @@
+import os
+import argparse
+from ete3 import Tree
+
+
+mcmctree_vs_reltime_usage = '''
+====================== mcmctree_vs_reltime example command ======================
+
+TreeSAK mcmctree_vs_reltime -m mcmc.out -r reltime.txt -n nodes.txt -o ages.pdf
+
+# Example data
+https://github.com/songweizhi/TreeSAK/tree/master/DemoData/mcmctree_vs_reltime
+
+=================================================================================
+'''
+
+
+def sep_path_basename_ext(file_in):
+
+    f_path, f_name = os.path.split(file_in)
+    if f_path == '':
+        f_path = '.'
+    f_base, f_ext = os.path.splitext(f_name)
+
+    return f_name, f_path, f_base, f_ext[1:]
+
+
+def get_lca(reltime_txt, leaf_1_name, leaf_2_name):
+
+    leaf_set = set()
+    child_to_parent_dict = dict()
+    id_to_name_dict = dict()
+    name_to_id_dict = dict()
+    for each_line in open(reltime_txt):
+        if not each_line.startswith('NodeLabel'):
+            each_line_split = each_line.strip().split('\t')
+            each_line_split = [i.strip() for i in each_line_split]
+            if len(each_line_split) > 1:
+                node_name = each_line_split[0].replace(' ', '_')
+                node_id = each_line_split[1]
+                des1 = each_line_split[2]
+                des2 = each_line_split[3]
+                id_to_name_dict[node_id] = node_name
+                name_to_id_dict[node_name] = node_id
+                child_to_parent_dict[des1] = node_id
+                child_to_parent_dict[des2] = node_id
+                if (des1 == '-') and (des2 == '-'):
+                    leaf_set.add(node_id)
+
+    leaf_to_lineage_dict = dict()
+    for leaf in sorted([i for i in leaf_set]):
+        original_leaf = leaf
+        lineage_list = [leaf]
+        while leaf in child_to_parent_dict:
+            leaf_p = child_to_parent_dict[leaf]
+            lineage_list.append(leaf_p)
+            leaf = leaf_p
+        leaf_to_lineage_dict[original_leaf] = lineage_list
+
+    leaf_1_id     = name_to_id_dict[leaf_1_name]
+    leaf_2_id     = name_to_id_dict[leaf_2_name]
+    leaf_1_linage = leaf_to_lineage_dict[leaf_1_id]
+    leaf_2_linage = leaf_to_lineage_dict[leaf_2_id]
+
+    lca = ''
+    for each_p in leaf_1_linage[::-1]:
+        if each_p in leaf_2_linage:
+            lca = each_p
+
+    return lca
+
+
+def parse_reltime(reltime_txt, interested_nodes_txt, op_txt):
+
+    lca_to_leaves_dict = dict()
+    interested_node_desc_dict = dict()
+    for interested_node in open(interested_nodes_txt):
+        interested_node_split = interested_node.strip().split('\t')
+        paired_leaves = interested_node_split[0]
+        interested_node_desc = paired_leaves
+        if len(interested_node_split) > 1:
+            interested_node_desc = interested_node_split[1]
+        interested_node_desc_dict[paired_leaves] = interested_node_desc
+        leaf_1 = paired_leaves.split(',')[0]
+        leaf_2 = paired_leaves.split(',')[1]
+        lca_id = get_lca(reltime_txt, leaf_1, leaf_2)
+        lca_to_leaves_dict[lca_id] = paired_leaves.strip()
+
+    op_txt_handle = open(op_txt, 'w')
+    line_num_index = 0
+    for each_line in open(reltime_txt):
+        each_line_split = each_line.strip().split('\t')
+        each_line_split = [i.strip() for i in each_line_split]
+        if line_num_index == 0:
+            op_txt_handle.write('ColorBy\tNode\tMean\tLow\tHigh\n')
+        else:
+            if len(each_line_split) > 1:
+                node_id = each_line_split[1]
+                if node_id in lca_to_leaves_dict:
+                    node_id = each_line_split[1]
+                    div_time = each_line_split[7]
+                    ci_lower = each_line_split[8]
+                    ci_upper = each_line_split[9]
+                    corresponding_leaves = lca_to_leaves_dict[node_id]
+                    interested_node_desc = interested_node_desc_dict[corresponding_leaves]
+                    op_txt_handle.write('RelTime\t%s\t%s\t%s\t%s\n' % (interested_node_desc, div_time, ci_lower, ci_upper))
+        line_num_index += 1
+    op_txt_handle.close()
+
+
+def mcmctree_out_to_tree_str(mamctree_out):
+
+    # get tree string from mamctree_out
+    tree_str = ''
+    tree_line = 0
+    current_line = 1
+    for each_line in open(mamctree_out):
+        if 'Species tree for FigTree.  Branch lengths = posterior mean times; 95% CIs = labels' in each_line:
+            tree_line = current_line + 1
+        if tree_line == current_line:
+            tree_str = each_line.strip()
+        current_line += 1
+
+    tree_str_no_space = tree_str.replace(' ', '')
+
+    # rename tree nodes
+    t = Tree(tree_str_no_space, format=1)
+    for each_node in t.traverse():
+        if each_node.is_leaf():
+            node_name_new = '_'.join(each_node.name.split('_')[1:])
+        else:
+            node_name_new = 't_n%s' % each_node.name
+        each_node.name = node_name_new
+
+    tree_str_renamed = t.write(format=8)
+
+    return tree_str_renamed
+
+
+def get_internal_node_to_plot(node_txt, mo_file):
+
+    tree_str = ''
+    if os.path.isfile(mo_file):
+        tree_str = mcmctree_out_to_tree_str(mo_file)
+
+    # get nodes to plot
+    node_set = set()
+    node_rename_dict = dict()
+    if os.path.isfile(node_txt) is True:
+        for each in open(node_txt):
+            each_split = each.strip().split('\t')
+            node_str = each_split[0]
+
+            # get internal_node_to_plot
+            internal_node_to_plot = ''
+            if ',' not in node_str:
+                internal_node_to_plot = each_split[0]
+            else:
+                leaf_list = node_str.split(',')
+                if tree_str == '':
+                    print('MCMCTree out file not found, program exited!')
+                    exit()
+                current_lca = Tree(tree_str, format=1).get_common_ancestor(leaf_list)
+                internal_node_to_plot = current_lca.name
+
+            # add internal_node_to_plot to  node_set
+            if internal_node_to_plot != '':
+                node_set.add(internal_node_to_plot)
+
+            # read in name to show in plot
+            if len(each_split) == 2:
+                if each_split[1] != '':
+                    node_rename_dict[internal_node_to_plot] = each_split[1]
+    else:
+        node_set = node_txt.split(',')
+
+    return node_set, node_rename_dict, tree_str
+
+
+def read_in_posterior_mean(mcmctree_out):
+
+    # read in Posterior mean
+    node_to_mean_hpd95_dict = dict()
+    current_line = 1
+    posterior_mean_header_line = 0
+    for each_line in open(mcmctree_out):
+        if 'Posterior mean (95% Equal-tail CI) (95% HPD CI) HPD-CI-width' in each_line:
+            posterior_mean_header_line = current_line
+
+        if (posterior_mean_header_line != 0) and (current_line > posterior_mean_header_line):
+            each_line_split = each_line.strip().split(' ')
+
+            each_line_split_no_empty = []
+            for each_element in each_line_split:
+                if each_element not in ['', '(']:
+                    each_element_value = each_element.replace('(', '').replace(')', '').replace(',', '')
+                    each_line_split_no_empty.append(each_element_value)
+            if len(each_line_split_no_empty) == 9:
+                node_id           = each_line_split_no_empty[0]
+                value_mean        = each_line_split_no_empty[1]
+                value_hpd95_small = each_line_split_no_empty[4]
+                value_hpd95_big   = each_line_split_no_empty[5]
+                node_to_mean_hpd95_dict[node_id] = [value_mean, value_hpd95_small, value_hpd95_big]
+        current_line += 1
+
+    return node_to_mean_hpd95_dict
+
+
+def parse_mcmc_out(mcmc_out_file, node_txt, dm_out):
+
+    dm_out_handle = open(dm_out, 'a')
+    #dm_out_handle.write('Test\tShape\tVar\tMean\tLow\tHigh\n')
+    node_set, node_rename_dict, tree_str = get_internal_node_to_plot(node_txt, mcmc_out_file)
+    node_to_mean_95_hpd_dict = read_in_posterior_mean(mcmc_out_file)
+    for each_node in node_set:
+        node_name_to_write = node_rename_dict.get(each_node, each_node)
+        mean_95_hpd_list = node_to_mean_95_hpd_dict.get(each_node)
+        dm_out_handle.write('MCMCTree\t%s\t%s\n' % (node_name_to_write, '\t'.join(mean_95_hpd_list)))
+    dm_out_handle.close()
+
+
+def mcmctree_vs_reltime(args):
+
+    mcmc_out_file           = args['m']
+    reltime_txt             = args['r']
+    interested_nodes_txt    = args['n']
+    pdf_out                 = args['o']
+
+    dm_out_combined = '.'.join(pdf_out.split('.')[:-1]) + '.txt'
+
+    # define mcmctree_vs_reltime_R
+    pwd_current_file  = os.path.realpath(__file__)
+    current_file_path = '/'.join(pwd_current_file.split('/')[:-1])
+    mcmctree_vs_reltime_R  = '%s/mcmctree_vs_reltime.R' % current_file_path
+
+    parse_reltime(reltime_txt, interested_nodes_txt, dm_out_combined)
+    parse_mcmc_out(mcmc_out_file, interested_nodes_txt, dm_out_combined)
+
+    plot_cmd   = 'Rscript %s -i %s -x %s -y %s -o %s' % (mcmctree_vs_reltime_R, dm_out_combined, 8, 5, pdf_out)
+    os.system(plot_cmd)
+    print('Plot exported to: %s' % pdf_out)
+
+
+if __name__ == '__main__':
+
+    mcmctree_vs_reltime_parser = argparse.ArgumentParser()
+    mcmctree_vs_reltime_parser.add_argument('-m',   required=True,  help='.out file from MCMCTree')
+    mcmctree_vs_reltime_parser.add_argument('-r',   required=True,  help='output from elTime')
+    mcmctree_vs_reltime_parser.add_argument('-n',   required=True,  help='interested nodes txt file')
+    mcmctree_vs_reltime_parser.add_argument('-o',   required=True,  help='output pdf')
+    args = vars(mcmctree_vs_reltime_parser.parse_args())
+    mcmctree_vs_reltime(args)

TreeSAK/merge_pdf.py

@@ -0,0 +1,32 @@
+import os
+import glob
+import math
+import random
+import argparse
+import seaborn as sns
+from ete3 import Tree
+from itolapi import Itol
+from PyPDF3.pdf import PageObject
+from PyPDF3 import PdfFileWriter, PdfFileReader
+from PyPDF3.generic import RectangleObject
+# https://pypdf2.readthedocs.io/en/3.0.0/user/adding-pdf-annotations.html
+
+def merge_pdf(pdf_1, pdf_2, margin_size, op_pdf):
+
+    page1 = PdfFileReader(open(pdf_1, "rb"), strict=False).getPage(0)
+    page2 = PdfFileReader(open(pdf_2, "rb"), strict=False).getPage(0)
+
+    total_width  = page1.mediaBox.upperRight[0] + page2.mediaBox.upperRight[0] + margin_size*3
+    total_height = max([page1.mediaBox.upperRight[1], page2.mediaBox.upperRight[1]]) + margin_size*2
+    new_page = PageObject.createBlankPage(None, total_width, total_height)
+    new_page.mergeTranslatedPage(page1, margin_size, (total_height-margin_size-page1.mediaBox.upperRight[1]))
+    new_page.mergeTranslatedPage(page2, (page1.mediaBox.upperRight[0] + margin_size*2), margin_size)
+    output = PdfFileWriter()
+    output.addPage(new_page)
+    output.write(open(op_pdf, "wb"))
+
+
+merge_pdf('/Users/songweizhi/Desktop/1.pdf',
+          '/Users/songweizhi/Desktop/2.pdf',
+          66,
+          '/Users/songweizhi/Desktop/merged.pdf')

TreeSAK/pRTC.py

@@ -0,0 +1,56 @@
+import os
+import argparse
+
+
+pRTC_usage = '''
+==================================== pRTC example commands ====================================
+
+# requires: ruby (bio-nwk, colorize, parallel, csv)
+
+TreeSAK pRTC -i out -m mcmc.txt -r rtc_dir -o after_pRTC_mcmc.txt
+
+python3 # format of RTC file (tab separated)
+symbiont_1,symbiont_2    host_1,host_2:0.9876
+recipient1,recipient2	 donor1,donor2:0.6789
+
+# The above two lines can be interpreted as follows:
+The probability of the last common ancestor (LCA) of symbiont_1 and symbiont_2 being 
+younger than that of host_1 and host_2 is 0.9876. Similarly, for a gene transfer event, 
+the LCA of the two recipients is younger than that of the two donors, the value 0.6789 
+is the PROBABILITY of the occurrence of the transfer event. Please do NOT use this module 
+if you are unsure about the interpretation of the value you provided.
+
+# Note
+This is a Python wrapper to perform the probabilistic RTC dating proposed by Dr. Sishuo Wang.
+If you used it in your dating analysis, please cite:
+https://doi.org/10.1101/2023.06.18.545440 or https://github.com/evolbeginner/rrtc.
+
+===============================================================================================
+'''
+
+
+def pRTC(args):
+
+    in_file     = args['i']
+    mcmc_txt    = args['m']
+    rrtc_dir    = args['r']
+    ruby_exe    = args['ruby']
+    op_txt      = args['o']
+
+    current_file_path = '/'.join(os.path.realpath(__file__).split('/')[:-1])
+    do_rrtc_rb        = '%s/do_rrtc.rb'                             % current_file_path
+    do_rrtc_cmd       = '%s %s --mcmctxt %s -i %s --rrtc %s > %s'   % (ruby_exe, do_rrtc_rb, mcmc_txt, in_file, rrtc_dir, op_txt)
+    print('Running: ' + do_rrtc_cmd)
+    os.system(do_rrtc_cmd)
+
+
+if __name__ == '__main__':
+
+    pRTC_parser = argparse.ArgumentParser(usage=pRTC_usage)
+    pRTC_parser.add_argument('-i',     required=True,                   help='the file "out" generated by MCMCTree')
+    pRTC_parser.add_argument('-m',     required=True,                   help='the file "mcmc.txt" generated by MCMCTree')
+    pRTC_parser.add_argument('-r',     required=True,                   help='the folder that contains all RTCs')
+    pRTC_parser.add_argument('-o',     required=True,                   help='output txt file')
+    pRTC_parser.add_argument('-ruby',  required=False, default='ruby',  help='path to ruby executable file, default: ruby')
+    args = vars(pRTC_parser.parse_args())
+    pRTC(args)