annotate fsd.py @ 19:b7bccbbee4a7 draft

planemo upload for repository https://github.com/monikaheinzl/duplexanalysis_galaxy/tree/master/tools/fsd commit b8a2f7b7615b2bcd3b602027af31f4e677da94f6-dirty

#!/usr/bin/env python

# Family size distribution of SSCSs
#
# Author: Monika Heinzl, Johannes-Kepler University Linz (Austria)
# Contact: [e-mail]
#
# Takes at least one TABULAR file with tags before the alignment to the SSCS, but up to 4 files can be provided, as input.
# The program produces a plot which shows the distribution of family sizes of the all SSCSs from the input files and
# a tabular file with the data of the plot, as well as a TXT file with all tags of the DCS and their family sizes.
# If only one file is provided, then a family size distribution, which is separated after SSCSs without a partner and DCSs, is produced.
# Whereas a family size distribution with multiple data in one plot is produced, when more than one file (up to 4) is given.

# USAGE: python FSD_Galaxy_1.4_commandLine_FINAL.py --inputFile1 filename --inputName1 filename --inputFile2 filename2 --inputName2 filename2 --inputFile3 filename3 --inputName3 filename3 --inputFile4 filename4 --inputName4 filename4 --log_axis --output_tabular outptufile_name_tabular --output_pdf outptufile_name_pdf

import argparse
import sys
import os

import matplotlib.pyplot as plt
import numpy
from matplotlib.backends.backend_pdf import PdfPages

plt.switch_backend('agg')


def readFileReferenceFree(file):
    with open(file, 'r') as dest_f:
        data_array = numpy.genfromtxt(dest_f, skip_header=0, delimiter='\t', comments='#', dtype='string')
        return(data_array)


def make_argparser():
    parser = argparse.ArgumentParser(description='Family Size Distribution of duplex sequencing data')
    parser.add_argument('--inputFile1', help='Tabular File with three columns: ab or ba, tag and family size.')
    parser.add_argument('--inputName1')
    parser.add_argument('--inputFile2', default=None, help='Tabular File with three columns: ab or ba, tag and family size.')
    parser.add_argument('--inputName2')
    parser.add_argument('--inputFile3', default=None, help='Tabular File with three columns: ab or ba, tag and family size.')
    parser.add_argument('--inputName3')
    parser.add_argument('--inputFile4', default=None, help='Tabular File with three columns: ab or ba, tag and family size.')
    parser.add_argument('--inputName4')
    parser.add_argument('--log_axis', action="store_false", help='Transform y axis in log scale.')
    parser.add_argument('--output_pdf', default="data.pdf", type=str, help='Name of the pdf file.')
    parser.add_argument('--output_tabular', default="data.tabular", type=str, help='Name of the tabular file.')
    return parser


def compare_read_families(argv):

    parser = make_argparser()
    args = parser.parse_args(argv[1:])

    firstFile = args.inputFile1
    name1 = args.inputName1

    secondFile = args.inputFile2
    name2 = args.inputName2
    thirdFile = args.inputFile3
    name3 = args.inputName3
    fourthFile = args.inputFile4
    name4 = args.inputName4
    log_axis = args.log_axis

    title_file = args.output_tabular
    title_file2 = args.output_pdf

    sep = "\t"

    plt.rc('figure', figsize=(11.69, 8.27))  # A4 format
    plt.rcParams['patch.edgecolor'] = "black"
    plt.rcParams['axes.facecolor'] = "E0E0E0"  # grey background color
    plt.rcParams['xtick.labelsize'] = 14
    plt.rcParams['ytick.labelsize'] = 14

    list_to_plot = []
    label = []
    data_array_list = []
    list_to_plot_original = []
    colors = []

    with open(title_file, "w") as output_file, PdfPages(title_file2) as pdf:
        fig = plt.figure()
        fig.subplots_adjust(left=0.12, right=0.97, bottom=0.23, top=0.94, hspace=0)
        fig2 = plt.figure()
        fig2.subplots_adjust(left=0.12, right=0.97, bottom=0.23, top=0.94, hspace=0)

        # plt.subplots_adjust(bottom=0.25)
        if firstFile != str(None):
            file1 = readFileReferenceFree(firstFile)
            integers = numpy.array(file1[:, 0]).astype(int)  # keep original family sizes
            list_to_plot_original.append(integers)
            colors.append("#0000FF")

            # for plot: replace all big family sizes by 22
            # data1 = numpy.array(file1[:, 0]).astype(int)
            # bigFamilies = numpy.where(data1 > 20)[0]
            # data1[bigFamilies] = 22
            if numpy.amax(integers) > 20:
                bins = numpy.arange(numpy.amin(integers), numpy.amax(integers) + 1)
                data1 = numpy.clip(integers, bins[0], bins[-1])
            else:
                data1 = integers
            name1 = name1.split(".tabular")[0]
            list_to_plot.append(data1)
            label.append(name1)
            data_array_list.append(file1)

            legend = "\n\n\n{}".format(name1)
            fig.text(0.05, 0.11, legend, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.05, 0.11, legend, size=10, transform=plt.gcf().transFigure)

            legend1 = "singletons:\nnr. of tags\n{:,} ({:.3f})".format(numpy.bincount(data1)[1], float(numpy.bincount(data1)[1]) / len(data1))
            fig.text(0.32, 0.11, legend1, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.32, 0.11, legend1, size=10, transform=plt.gcf().transFigure)

            legend3b = "PE reads\n{:,} ({:.3f})".format(numpy.bincount(data1)[1], float(numpy.bincount(data1)[1]) / sum(integers))
            fig.text(0.45, 0.11, legend3b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.45, 0.11, legend3b, size=10, transform=plt.gcf().transFigure)

            legend4 = "family size > 20:\nnr. of tags\n{:,} ({:.3f})".format(numpy.bincount(data1)[len(numpy.bincount(data1)) - 1].astype(int), float(numpy.bincount(data1)[len(numpy.bincount(data1)) - 1]) / len(data1))
            fig.text(0.58, 0.11, legend4, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.58, 0.11, legend4, size=10, transform=plt.gcf().transFigure)

            legend5 = "PE reads\n{:,} ({:.3f})".format(sum(integers[integers > 20]), float(sum(integers[integers > 20])) / sum(integers))
            fig.text(0.70, 0.11, legend5, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.70, 0.11, legend5, size=10, transform=plt.gcf().transFigure)

            legend6 = "total nr. of\ntags\n{:,}".format(len(data1))
            fig.text(0.82, 0.11, legend6, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.82, 0.11, legend6, size=10, transform=plt.gcf().transFigure)

            legend6b = "PE reads\n{:,}".format(sum(integers))
            fig.text(0.89, 0.11, legend6b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.89, 0.11, legend6b, size=10, transform=plt.gcf().transFigure)

        if secondFile != str(None):
            file2 = readFileReferenceFree(secondFile)
            integers2 = numpy.array(file2[:, 0]).astype(int)  # keep original family sizes
            list_to_plot_original.append(integers2)
            colors.append("#298A08")

            # data2 = numpy.asarray(file2[:, 0]).astype(int)
            # bigFamilies2 = numpy.where(data2 > 20)[0]
            # data2[bigFamilies2] = 22

            if numpy.amax(integers) > 20:
                bins = numpy.arange(numpy.amin(integers2), numpy.amax(integers2) + 1)
                data2 = numpy.clip(integers2, bins[0], bins[-1])
            else:
                data2 = integers2
            list_to_plot.append(data2)
            name2 = name2.split(".tabular")[0]
            label.append(name2)
            data_array_list.append(file2)

            fig.text(0.05, 0.09, name2, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.05, 0.09, name2, size=10, transform=plt.gcf().transFigure)

            legend1 = "{:,} ({:.3f})".format(numpy.bincount(data2)[1], float(numpy.bincount(data2)[1]) / len(data2))
            fig.text(0.32, 0.09, legend1, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.32, 0.09, legend1, size=10, transform=plt.gcf().transFigure)

            legend3 = "{:,} ({:.3f})".format(numpy.bincount(data2)[1], float(numpy.bincount(data2)[1]) / sum(integers2))
            fig.text(0.45, 0.09, legend3, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.45, 0.09, legend3, size=10, transform=plt.gcf().transFigure)

            legend4 = "{:,} ({:.3f})".format(
                numpy.bincount(data2)[len(numpy.bincount(data2)) - 1].astype(int),
                float(numpy.bincount(data2)[len(numpy.bincount(data2)) - 1]) / len(data2))
            fig.text(0.58, 0.09, legend4, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.58, 0.09, legend4, size=10, transform=plt.gcf().transFigure)

            legend5 = "{:,} ({:.3f})".format(sum(integers2[integers2 > 20]), float(sum(integers2[integers2 > 20])) / sum(integers2))
            fig.text(0.70, 0.09, legend5, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.70, 0.09, legend5, size=10, transform=plt.gcf().transFigure)

            legend6 = "{:,}".format(len(data2))
            fig.text(0.82, 0.09, legend6, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.82, 0.09, legend6, size=10, transform=plt.gcf().transFigure)

            legend6b = "{:,}".format(sum(integers2))
            fig.text(0.89, 0.09, legend6b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.89, 0.09, legend6b, size=10, transform=plt.gcf().transFigure)

        if thirdFile != str(None):
            file3 = readFileReferenceFree(thirdFile)
            integers3 = numpy.array(file3[:, 0]).astype(int)  # keep original family sizes
            list_to_plot_original.append(integers3)
            colors.append("#DF0101")

            # data3 = numpy.asarray(file3[:, 0]).astype(int)
            # bigFamilies3 = numpy.where(data3 > 20)[0]
            # data3[bigFamilies3] = 22

            if numpy.amax(integers3) > 20:
                bins = numpy.arange(numpy.amin(integers3), numpy.amax(integers3) + 1)
                data3 = numpy.clip(integers3, bins[0], bins[-1])
            else:
                data3 = integers3
            list_to_plot.append(data3)
            name3 = name3.split(".tabular")[0]
            label.append(name3)
            data_array_list.append(file3)

            fig.text(0.05, 0.07, name3, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.05, 0.07, name3, size=10, transform=plt.gcf().transFigure)

            legend1 = "{:,} ({:.3f})".format(numpy.bincount(data3)[1], float(numpy.bincount(data3)[1]) / len(data3))
            fig.text(0.32, 0.07, legend1, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.32, 0.07, legend1, size=10, transform=plt.gcf().transFigure)

            legend3b = "{:,} ({:.3f})".format(numpy.bincount(data3)[1], float(numpy.bincount(data3)[1]) / sum(integers3))
            fig.text(0.45, 0.07, legend3b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.45, 0.07, legend3b, size=10, transform=plt.gcf().transFigure)

            legend4 = "{:,} ({:.3f})".format(
                numpy.bincount(data3)[len(numpy.bincount(data3)) - 1].astype(int),
                float(numpy.bincount(data3)[len(numpy.bincount(data3)) - 1]) / len(data3))
            fig.text(0.58, 0.07, legend4, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.58, 0.07, legend4, size=10, transform=plt.gcf().transFigure)

            legend5 = "{:,} ({:.3f})".format(sum(integers3[integers3 > 20]),
                                             float(sum(integers3[integers3 > 20])) / sum(integers3))
            fig.text(0.70, 0.07, legend5, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.70, 0.07, legend5, size=10, transform=plt.gcf().transFigure)

            legend6 = "{:,}".format(len(data3))
            fig.text(0.82, 0.07, legend6, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.82, 0.07, legend6, size=10, transform=plt.gcf().transFigure)

            legend6b = "{:,}".format(sum(integers3))
            fig.text(0.89, 0.07, legend6b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.89, 0.07, legend6b, size=10, transform=plt.gcf().transFigure)

        if fourthFile != str(None):
            file4 = readFileReferenceFree(fourthFile)
            integers4 = numpy.array(file4[:, 0]).astype(int)  # keep original family sizes
            list_to_plot_original.append(integers4)
            colors.append("#04cec7")

            # data4 = numpy.asarray(file4[:, 0]).astype(int)
            # bigFamilies4 = numpy.where(data4 > 20)[0]
            # data4[bigFamilies4] = 22
            if numpy.amax(integers4) > 20:
                bins = numpy.arange(numpy.amin(integers4), numpy.amax(integers4) + 1)
                data4 = numpy.clip(integers4, bins[0], bins[-1])
            else:
                data4 = integers4
            list_to_plot.append(data4)
            name4 = name4.split(".tabular")[0]
            label.append(name4)
            data_array_list.append(file4)

            fig.text(0.05, 0.05, name4, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.05, 0.05, name4, size=10, transform=plt.gcf().transFigure)

            legend1 = "{:,} ({:.3f})".format(numpy.bincount(data4)[1], float(numpy.bincount(data4)[1]) / len(data4))
            fig.text(0.32, 0.05, legend1, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.32, 0.05, legend1, size=10, transform=plt.gcf().transFigure)

            legend3b = "{:,} ({:.3f})".format(numpy.bincount(data4)[1], float(numpy.bincount(data4)[1]) / sum(integers4))
            fig.text(0.45, 0.05, legend3b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.45, 0.05, legend3b, size=10, transform=plt.gcf().transFigure)

            legend4 = "{:,} ({:.3f})".format(
                numpy.bincount(data4)[len(numpy.bincount(data4)) - 1].astype(int),
                float(numpy.bincount(data4)[len(numpy.bincount(data4)) - 1]) / len(data4))
            fig.text(0.58, 0.05, legend4, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.58, 0.05, legend4, size=10, transform=plt.gcf().transFigure)

            legend5 = "{:,} ({:.3f})".format(sum(integers4[integers4 > 20]),
                                             float(sum(integers4[integers4 > 20])) / sum(integers4))
            fig.text(0.70, 0.05, legend5, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.70, 0.05, legend5, size=10, transform=plt.gcf().transFigure)

            legend6 = "{:,}".format(len(data4))
            fig.text(0.82, 0.05, legend6, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.82, 0.05, legend6, size=10, transform=plt.gcf().transFigure)

            legend6b = "{:,}".format(sum(integers4))
            fig.text(0.89, 0.05, legend6b, size=10, transform=plt.gcf().transFigure)
            fig2.text(0.89, 0.05, legend6b, size=10, transform=plt.gcf().transFigure)

        maximumX = numpy.amax(numpy.concatenate(list_to_plot))
        minimumX = numpy.amin(numpy.concatenate(list_to_plot))
        bins = numpy.arange(minimumX, maximumX + 1)
        list_to_plot2 = list_to_plot
        to_plot = ["Absolute frequencies", "Relative frequencies"]
        plt.xticks([], [])
        plt.yticks([], [])
        fig.suptitle('Family Size Distribution (tags)', fontsize=14)

        for l in range(len(to_plot)):
            ax = fig.add_subplot(2, 1, l+1)
            ticks = numpy.arange(1, 22, 1)
            ticks1 = map(str, ticks)
            if maximumX > 20:
                ticks1[len(ticks1) - 1] = ">20"

            if to_plot[l] == "Relative frequencies":
                counts_rel = ax.hist(list_to_plot2, bins=numpy.arange(minimumX, maximumX + 2), stacked=False, edgecolor="black", linewidth=1, label=label, align="left", alpha=1, rwidth=0.8, normed=True)
            else:
                counts = ax.hist(list_to_plot2, bins=numpy.arange(minimumX, maximumX + 2), stacked=False, edgecolor="black", linewidth=1, label=label, align="left", alpha=1, rwidth=0.8)
                ax.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(0.9, 1))

            ax.set_xticks(numpy.array(ticks))
            ax.set_xticklabels(ticks1)

            ax.set_ylabel(to_plot[l], fontsize=14)
            ax.set_xlabel("Family size", fontsize=14)
            if log_axis:
                ax.set_yscale('log')
            ax.grid(b=True, which="major", color="#424242", linestyle=":")
            ax.margins(0.01, None)
        pdf.savefig(fig)
        plt.close()

        fig2.suptitle('Family Size Distribution (PE reads)', fontsize=14)
        for l in range(len(to_plot)):
            ax = fig2.add_subplot(2, 1, l + 1)
            ticks = numpy.arange(minimumX, maximumX + 1)
            ticks1 = map(str, ticks)
            if maximumX > 20:
                ticks1[len(ticks1) - 1] = ">20"
            reads = []
            reads_rel = []

            barWidth = 0 - (len(list_to_plot)+1)/2 * 1./(len(list_to_plot) + 1)

            for i in range(len(list_to_plot2)):
                unique, c = numpy.unique(list_to_plot2[i], return_counts=True)
                new_c = []
                new_unique = []
                
                for t in ticks:
                    if t not in unique:
                        new_c.append(0) # add zero count of not occuring
                        new_unique.append(t)
                    else:
                        c_idx = numpy.where(t == unique)[0]
                        new_c.append(c[c_idx])
                        new_unique.append(unique[c_idx])
                print(new_unique)
                print(new_c)
                y = numpy.array(new_unique) * numpy.array(new_c)
                if len([list_to_plot_original > 20]) > 0:
                    y[len(y) - 1] = sum(list_to_plot_original[i][list_to_plot_original[i] > 20])
                reads.append(y)
                reads_rel.append(list(numpy.float_(y)) / sum(y))

                x = list(numpy.arange(numpy.amin(unique), numpy.amax(unique) + 1).astype(float))
                x = [xi + barWidth for xi in x]

                if to_plot[l] == "Relative frequencies":
                    counts2_rel = ax.bar(x, list(numpy.float_(y)) / sum(y), align="edge", width=1./(len(list_to_plot) + 1),
                                         edgecolor="black", label=label[i], alpha=1, linewidth=1, color=colors[i])
                else:
                    counts2 = ax.bar(x, y, align="edge", width=1./len(list_to_plot), edgecolor="black", label=label[i],
                                     alpha=1, linewidth=1, color=colors[i])
                if i == len(list_to_plot2):
                    barWidth += 1. / (len(list_to_plot) + 1) + 1. / (len(list_to_plot) + 1)
                else:
                    barWidth += 1. / (len(list_to_plot) + 1)

            if to_plot[l] == "Absolute frequencies":
                ax.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(0.9, 1))
            else:
                ax.set_xlabel("Family size", fontsize=14)

            ax.set_xticks(numpy.array(ticks))
            ax.set_xticklabels(ticks1)
            ax.set_ylabel(to_plot[l], fontsize=14)
            if log_axis:
                ax.set_yscale('log')
            ax.grid(b=True, which="major", color="#424242", linestyle=":")
            ax.margins(0.01, None)

        pdf.savefig(fig2)
        plt.close()

        # write data to CSV file tags
        output_file.write("Values from family size distribution with all datasets (tags)\n")
        output_file.write("\nFamily size")
        for i in label:
            output_file.write("{}{}".format(sep, i))
        # output_file.write("{}sum".format(sep))
        output_file.write("\n")
        j = 0
        for fs in counts[1][0:len(counts[1]) - 1]:
            if fs == 21:
                fs = ">20"
            else:
                fs = "={}".format(fs)
            output_file.write("FS{}{}".format(fs, sep))
            if len(label) == 1:
                output_file.write("{}{}".format(int(counts[0][j]), sep))
            else:
                for n in range(len(label)):
                    output_file.write("{}{}".format(int(counts[0][n][j]), sep))
            output_file.write("\n")
            j += 1
        output_file.write("sum{}".format(sep))
        if len(label) == 1:
            output_file.write("{}{}".format(int(sum(counts[0])), sep))
        else:
            for i in counts[0]:
                output_file.write("{}{}".format(int(sum(i)), sep))

        # write data to CSV file PE reads
        output_file.write("\n\nValues from family size distribution with all datasets (PE reads)\n")
        output_file.write("\nFamily size")
        for i in label:
            output_file.write("{}{}".format(sep, i))
        # output_file.write("{}sum".format(sep))
        output_file.write("\n")
        j = 0
        for fs in bins:
            if fs == 21:
                fs = ">20"
            else:
                fs = "={}".format(fs)
            output_file.write("FS{}{}".format(fs, sep))
            if len(label) == 1:
                output_file.write("{}{}".format(int(reads[0][j]), sep))
            else:
                for n in range(len(label)):
                    output_file.write("{}{}".format(int(reads[n][j]), sep))
            output_file.write("\n")
            j += 1
        output_file.write("sum{}".format(sep))
        if len(label) == 1:
            output_file.write("{}{}".format(int(sum(reads)), sep))
        else:
            for i in reads:
                output_file.write("{}{}".format(int(sum(i)), sep))
        output_file.write("\n")

        # Family size distribution after DCS and SSCS
        for dataset, data_o, name_file in zip(list_to_plot, data_array_list, label):
            maximumX = numpy.amax(dataset)
            minimumX = numpy.amin(dataset)

            tags = numpy.array(data_o[:, 2])
            seq = numpy.array(data_o[:, 1])
            data = numpy.array(dataset)
            data_o = numpy.array(data_o[:, 0]).astype(int)
            # find all unique tags and get the indices for ALL tags, but only once
            u, index_unique, c = numpy.unique(numpy.array(seq), return_counts=True, return_index=True)
            d = u[c > 1]

            # get family sizes, tag for duplicates
            duplTags_double = data[numpy.in1d(seq, d)]
            duplTags_double_o = data_o[numpy.in1d(seq, d)]

            duplTags = duplTags_double[0::2]  # ab of DCS
            duplTags_o = duplTags_double_o[0::2]  # ab of DCS

            duplTagsBA = duplTags_double[1::2]  # ba of DCS
            duplTagsBA_o = duplTags_double_o[1::2]  # ba of DCS

            # duplTags_double_tag = tags[numpy.in1d(seq, d)]
            # duplTags_double_seq = seq[numpy.in1d(seq, d)]

            # get family sizes for SSCS with no partner
            ab = numpy.where(tags == "ab")[0]
            abSeq = seq[ab]
            ab_o = data_o[ab]
            ab = data[ab]

            ba = numpy.where(tags == "ba")[0]
            baSeq = seq[ba]
            ba_o = data_o[ba]
            ba = data[ba]

            dataAB = ab[numpy.in1d(abSeq, d, invert=True)]
            dataAB_o = ab_o[numpy.in1d(abSeq, d, invert=True)]

            dataBA = ba[numpy.in1d(baSeq, d, invert=True)]
            dataBA_o = ba_o[numpy.in1d(baSeq, d, invert=True)]

            list1 = [duplTags_double, dataAB, dataBA]  # list for plotting

            # information for family size >= 3
            dataAB_FS3 = dataAB[dataAB >= 3]
            dataAB_FS3_o = dataAB_o[dataAB_o >= 3]
            dataBA_FS3 = dataBA[dataBA >= 3]
            dataBA_FS3_o = dataBA_o[dataBA_o >= 3]
            # ab_FS3 = ab[ab >= 3]
            # ba_FS3 = ba[ba >= 3]
            # ab_FS3_o = ab_o[ab_o >= 3]
            # ba_FS3_o = ba_o[ba_o >= 3]

            duplTags_FS3 = duplTags[(duplTags >= 3) & (duplTagsBA >= 3)]  # ab+ba with FS>=3
            duplTags_FS3_BA = duplTagsBA[(duplTags >= 3) & (duplTagsBA >= 3)]  # ba+ab with FS>=3
            duplTags_double_FS3 = len(duplTags_FS3) + len(duplTags_FS3_BA)  # both ab and ba strands with FS>=3

            # original FS
            duplTags_FS3_o = duplTags_o[(duplTags_o >= 3) & (duplTagsBA_o >= 3)]  # ab+ba with FS>=3
            duplTags_FS3_BA_o = duplTagsBA_o[(duplTags_o >= 3) & (duplTagsBA_o >= 3)]  # ba+ab with FS>=3
            duplTags_double_FS3_o = sum(duplTags_FS3_o) + sum(duplTags_FS3_BA_o)  # both ab and ba strands with FS>=3

            fig = plt.figure()
            plt.subplots_adjust(left=0.12, right=0.97, bottom=0.3, top=0.94, hspace=0)
            counts = plt.hist(list1, bins=numpy.arange(minimumX, maximumX + 2), stacked=True, label=["duplex", "ab", "ba"],
                              edgecolor="black", linewidth=1, align="left", color=["#FF0000", "#5FB404", "#FFBF00"],
                              rwidth=0.8)
            # tick labels of x axis
            ticks = numpy.arange(1, 22, 1)
            ticks1 = map(str, ticks)
            if maximumX > 20:
                ticks1[len(ticks1) - 1] = ">20"
            plt.xticks(numpy.array(ticks), ticks1)
            singl = counts[0][2][0]  # singletons
            last = counts[0][2][len(counts[0][0]) - 1]  # large families
            if log_axis:
                plt.yscale('log')
            plt.legend(loc='upper right', fontsize=14, bbox_to_anchor=(0.9, 1), frameon=True)
            plt.title(name_file, fontsize=14)
            plt.xlabel("Family size", fontsize=14)
            plt.ylabel("Absolute Frequency", fontsize=14)
            plt.margins(0.01, None)
            plt.grid(b=True, which="major", color="#424242", linestyle=":")

            # extra information beneath the plot
            legend = "SSCS ab= \nSSCS ba= \nDCS (total)= \ntotal nr. of tags="
            plt.text(0.1, 0.09, legend, size=10, transform=plt.gcf().transFigure)

            legend = "nr. of tags\n\n{:,}\n{:,}\n{:,} ({:,})\n{:,}".format(len(dataAB), len(dataBA), len(duplTags), len(duplTags_double), (len(dataAB) + len(dataBA) + len(duplTags)))
            plt.text(0.23, 0.09, legend, size=10, transform=plt.gcf().transFigure)

            legend5 = "PE reads\n\n{:,}\n{:,}\n{:,} ({:,})\n{:,}".format(sum(dataAB_o), sum(dataBA_o), sum(duplTags_o), sum(duplTags_double_o), (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)))
            plt.text(0.38, 0.09, legend5, size=10, transform=plt.gcf().transFigure)

            legend = "rel. freq. of tags\nunique\n{:.3f}\n{:.3f}\n{:.3f}\n{:,}".format(float(len(dataAB)) / (len(dataAB) + len(dataBA) + len(duplTags)), float(len(dataBA)) / (len(dataAB) + len(dataBA) + len(duplTags)), float(len(duplTags)) / (len(dataAB) + len(dataBA) + len(duplTags)), (len(dataAB) + len(dataBA) + len(duplTags)))
            plt.text(0.54, 0.09, legend, size=10, transform=plt.gcf().transFigure)

            legend = "total\n{:.3f}\n{:.3f}\n{:.3f} ({:.3f})\n{:,}".format(float(len(dataAB)) / (len(ab) + len(ba)), float(len(dataBA)) / (len(ab) + len(ba)), float(len(duplTags)) / (len(ab) + len(ba)), float(len(duplTags_double)) / (len(ab) + len(ba)), (len(ab) + len(ba)))
            plt.text(0.64, 0.09, legend, size=10, transform=plt.gcf().transFigure)

            legend1 = "\nsingletons:\nfamily size > 20:"
            plt.text(0.1, 0.03, legend1, size=10, transform=plt.gcf().transFigure)

            legend4 = "{:,}\n{:,}".format(singl.astype(int), last.astype(int))
            plt.text(0.23, 0.03, legend4, size=10, transform=plt.gcf().transFigure)

            legend3 = "{:.3f}\n{:.3f}".format(singl / len(data), last / len(data))
            plt.text(0.64, 0.03, legend3, size=10, transform=plt.gcf().transFigure)

            legend3 = "\n\n{:,}".format(sum(data_o[data_o > 20]))
            plt.text(0.38, 0.03, legend3, size=10, transform=plt.gcf().transFigure)

            legend3 = "{:.3f}\n{:.3f}".format(float(singl)/sum(data_o), float(sum(data_o[data_o > 20])) / sum(data_o))
            plt.text(0.84, 0.03, legend3, size=10, transform=plt.gcf().transFigure)

            legend = "PE reads\nunique\n{:.3f}\n{:.3f}\n{:.3f}\n{:,}".format(
                float(sum(dataAB_o)) / (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)),
                float(sum(dataBA_o)) / (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)),
                float(sum(duplTags_o)) / (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)),
                (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)))
            plt.text(0.74, 0.09, legend, size=10, transform=plt.gcf().transFigure)

            legend = "total\n{:.3f}\n{:.3f}\n{:.3f} ({:.3f})\n{:,}".format(
                float(sum(dataAB_o)) / (sum(ab_o) + sum(ba_o)),
                float(sum(dataBA_o)) / (sum(ab_o) + sum(ba_o)),
                float(sum(duplTags_o)) / (sum(ab_o) + sum(ba_o)),
                float(sum(duplTags_double_o)) / (sum(ab_o) + sum(ba_o)), (sum(ab_o) + sum(ba_o)))
            plt.text(0.84, 0.09, legend, size=10, transform=plt.gcf().transFigure)

            pdf.savefig(fig)
            plt.close()

            # write same information to a csv file
            count = numpy.bincount(integers)  # original counts of family sizes
            output_file.write("\nDataset:{}{}\n".format(sep, name_file))
            output_file.write("max. family size:{}{}\n".format(sep, max(integers)))
            output_file.write("absolute frequency:{}{}\n".format(sep, count[len(count) - 1]))
            output_file.write("relative frequency:{}{:.3f}\n\n".format(sep, float(count[len(count) - 1]) / sum(count)))

            output_file.write("{}singletons:{}{}{}family size > 20:{}{}{}{}length of dataset:\n".format(sep, sep, sep, sep, sep, sep, sep, sep))
            output_file.write("{}nr. of tags{}rel. freq of tags{}rel.freq of PE reads{}nr. of tags{}rel. freq of tags{}nr. of PE reads{}rel. freq of PE reads{}total nr. of tags{}total nr. of PE reads\n".format(sep, sep, sep, sep, sep, sep, sep, sep, sep))
            output_file.write("{}{}{}{}{:.3f}{}{:.3f}{}{}{}{:.3f}{}{}{}{:.3f}{}{}{}{}\n\n".format(
                name_file, sep, singl.astype(int), sep, singl / len(data), sep, float(singl)/sum(data_o), sep,
                last.astype(int), sep, last / len(data), sep, sum(data_o[data_o > 20]), sep, float(sum(data_o[data_o > 20])) / sum(data_o), sep, len(data), sep, sum(data_o)))

            # information for FS >= 1
            output_file.write("The unique frequencies were calculated from the dataset where the tags occured only once (=ab without DCS, ba without DCS)\n"
                              "Whereas the total frequencies were calculated from the whole dataset (=including the DCS).\n\n")
            output_file.write("FS >= 1{}nr. of tags{}nr. of PE reads{}rel. freq of tags{}{}rel. freq of PE reads:\n".format(sep, sep, sep, sep, sep))
            output_file.write("{}{}{}unique:{}total{}unique{}total:\n".format(sep, sep, sep, sep, sep, sep))
            output_file.write("SSCS ab{}{}{}{}{}{:.3f}{}{:.3f}{}{:.3f}{}{:.3f}\n".format(
                sep, len(dataAB), sep, sum(dataAB_o), sep, float(len(dataAB)) / (len(dataAB) + len(dataBA) + len(duplTags)),
                sep, float(sum(dataAB_o)) / (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)), sep,
                float(len(dataAB)) / (len(ab) + len(ba)), sep, float(sum(dataAB_o)) / (sum(ab_o) + sum(ba_o))))
            output_file.write("SSCS ba{}{}{}{}{}{:.3f}{}{:.3f}{}{:.3f}{}{:.3f}\n".format(
                sep, len(dataBA), sep, sum(dataBA_o), sep, float(len(dataBA)) / (len(dataBA) + len(dataBA) + len(duplTags)),
                sep, float(sum(dataBA_o)) / (sum(dataBA_o) + sum(dataBA_o) + sum(duplTags_o)), sep, float(len(dataBA)) / (len(ba) + len(ba)),
                sep, float(sum(dataBA_o)) / (sum(ba_o) + sum(ba_o))))
            output_file.write("DCS (total){}{} ({}){}{} ({}){}{:.3f}{}{:.3f} ({:.3f}){}{:.3f}{}{:.3f} ({:.3f})\n".format(
                sep, len(duplTags), len(duplTags_double), sep, sum(duplTags_o), sum(duplTags_double_o), sep,
                float(len(duplTags)) / (len(dataAB) + len(dataBA) + len(duplTags)), sep,
                float(len(duplTags)) / (len(ab) + len(ba)), float(len(duplTags_double)) / (len(ab) + len(ba)), sep,
                float(sum(duplTags_o)) / (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)), sep,
                float(sum(duplTags_o)) / (sum(ab_o) + sum(ba_o)), float(sum(duplTags_double_o)) / (sum(ab_o) + sum(ba_o))))
            output_file.write("total nr. of tags{}{}{}{}{}{}{}{}{}{}{}{}\n".format(
                sep, (len(dataAB) + len(dataBA) + len(duplTags)), sep, (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)), sep,
                (len(dataAB) + len(dataBA) + len(duplTags)), sep, (len(ab) + len(ba)), sep,
                (sum(dataAB_o) + sum(dataBA_o) + sum(duplTags_o)), sep, (sum(ab_o) + sum(ba_o))))
            # information for FS >= 3
            output_file.write("\nFS >= 3{}nr. of tags{}nr. of PE reads{}rel. freq of tags{}{}rel. freq of PE reads:\n".format(sep, sep, sep, sep, sep))
            output_file.write("{}{}{}unique:{}total{}unique{}total:\n".format(sep, sep, sep, sep, sep, sep))
            output_file.write("SSCS ab{}{}{}{}{}{:.3f}{}{:.3f}{}{:.3f}{}{:.3f}\n".format(
                sep, len(dataAB_FS3), sep, sum(dataAB_FS3_o), sep,
                float(len(dataAB_FS3)) / (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)), sep,
                float(len(dataAB_FS3)) / (len(dataBA_FS3) + len(dataBA_FS3) + duplTags_double_FS3),
                sep, float(sum(dataAB_FS3_o)) / (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + sum(duplTags_FS3_o)),
                sep, float(sum(dataAB_FS3_o)) / (sum(dataBA_FS3_o) + sum(dataBA_FS3_o) + duplTags_double_FS3_o)))
            output_file.write("SSCS ba{}{}{}{}{}{:.3f}{}{:.3f}{}{:.3f}{}{:.3f}\n".format(
                sep, len(dataBA_FS3), sep, sum(dataBA_FS3_o), sep,
                float(len(dataBA_FS3)) / (len(dataBA_FS3) + len(dataBA_FS3) + len(duplTags_FS3)),
                sep, float(len(dataBA_FS3)) / (len(dataBA_FS3) + len(dataBA_FS3) + duplTags_double_FS3),
                sep, float(sum(dataBA_FS3_o)) / (sum(dataBA_FS3_o) + sum(dataBA_FS3_o) + sum(duplTags_FS3_o)),
                sep, float(sum(dataBA_FS3_o)) / (sum(dataBA_FS3_o) + sum(dataBA_FS3_o) + duplTags_double_FS3_o)))
            output_file.write("DCS (total){}{} ({}){}{} ({}){}{:.3f}{}{:.3f} ({:.3f}){}{:.3f}{}{:.3f} ({:.3f})\n".format(
                sep, len(duplTags_FS3), duplTags_double_FS3,  sep, sum(duplTags_FS3_o), duplTags_double_FS3_o, sep,
                float(len(duplTags_FS3)) / (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)), sep,
                float(len(duplTags_FS3)) / (len(dataAB_FS3) + len(dataBA_FS3) + duplTags_double_FS3),
                float(duplTags_double_FS3) / (len(dataAB_FS3) + len(dataBA_FS3) + duplTags_double_FS3),
                sep, float(sum(duplTags_FS3_o)) / (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + sum(duplTags_FS3_o)), sep,
                float(sum(duplTags_FS3_o)) / (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + duplTags_double_FS3_o),
                float(duplTags_double_FS3_o) / (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + duplTags_double_FS3_o)))
            output_file.write("total nr. of tags{}{}{}{}{}{}{}{}{}{}{}{}\n".format(
                sep, (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)), sep, (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + sum(duplTags_FS3_o)),
                sep, (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)), sep, (len(dataAB_FS3) + len(dataBA_FS3) + duplTags_double_FS3),
                sep, (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + sum(duplTags_FS3_o)), sep, (sum(dataAB_FS3_o) + sum(dataBA_FS3_o) + duplTags_double_FS3_o)))

            output_file.write("\nValues from family size distribution\n")
            output_file.write("{}duplex{}ab{}ba{}sum\n".format(sep, sep, sep, sep))
            for dx, ab, ba, fs in zip(counts[0][0], counts[0][1], counts[0][2], counts[1]):
                if fs == 21:
                    fs = ">20"
                else:
                    fs = "={}".format(fs)
                ab1 = ab - dx
                ba1 = ba - ab
                output_file.write("FS{}{}{}{}{}{}{}{}{}\n".format(fs, sep, int(dx), sep, int(ab1), sep, int(ba1), sep, int(ba)))

    print("Files successfully created!")


if __name__ == '__main__':
    sys.exit(compare_read_families(sys.argv))