fsd: fsd.py annotate

annotate fsd.py @ 8:7d5c86d59850 draft

planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 72152f9651d7b1ecc84052f967d2746813481af6

author	mheinzl
date	Wed, 09 May 2018 09:35:23 -0400
parents	43b16247c35a
children	f7136e93604b

rev	line source
0 9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	1 #!/usr/bin/env python
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	2
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	3 # Family size distribution of SSCSs
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	4 #
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	5 # Author: Monika Heinzl, Johannes-Kepler University Linz (Austria)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	6 # Contact: monika.heinzl@edumail.at
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	7 #
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	8 # Takes at least one TABULAR file with tags before the alignment to the SSCS, but up to 4 files can be provided, as input.
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	9 # The program produces a plot which shows the distribution of family sizes of the all SSCSs from the input files and
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	10 # a CSV file with the data of the plot, as well as a TXT file with all tags of the DCS and their family sizes.
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	11 # If only one file is provided, then a family size distribution, which is separated after SSCSs without a partner and DCSs, is produced.
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	12 # Whereas a family size distribution with multiple data in one plot is produced, when more than one file (up to 4) is given.
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	13
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	14 # USAGE: python FSD_Galaxy_1.4_commandLine_FINAL.py filename --inputFile2 filename2 --inputFile3 filename3 --inputFile4 filename4 /
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	15 # --title_file outputFileName --sep "characterWhichSeparatesCSVFile"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	16
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	17 import numpy
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	18 import matplotlib.pyplot as plt
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	19 from matplotlib.backends.backend_pdf import PdfPages
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	20 import argparse
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	21 import sys
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	22 import os
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	23 import re
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	24
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	25 def readFileReferenceFree(file):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	26 with open(file, 'r') as dest_f:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	27 data_array = numpy.genfromtxt(dest_f, skip_header=0, delimiter='\t', comments='#', dtype='string')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	28 return(data_array)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	29
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	30 def make_argparser():
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	31 parser = argparse.ArgumentParser(description='Family Size Distribution of duplex sequencing data')
6 c4b8222dce29 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 1667d50b521325eea1eada1b16029b1ba1379c5f mheinzl parents: 5 diff changeset	32 parser.add_argument('--inputFile1',
c4b8222dce29 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 1667d50b521325eea1eada1b16029b1ba1379c5f mheinzl parents: 5 diff changeset	33 help='Tabular File with three columns: ab or ba, tag and family size.')
0 9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	34 parser.add_argument('--inputName1')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	35 parser.add_argument('--inputFile2',default=None,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	36 help='Tabular File with three columns: ab or ba, tag and family size.')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	37 parser.add_argument('--inputName2')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	38 parser.add_argument('--inputFile3',default=None,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	39 help='Tabular File with three columns: ab or ba, tag and family size.')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	40 parser.add_argument('--inputName3')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	41 parser.add_argument('--inputFile4',default=None,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	42 help='Tabular File with three columns: ab or ba, tag and family size.')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	43 parser.add_argument('--inputName4')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	44 parser.add_argument('--sep', default=",",
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	45 help='Separator in the csv file.')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	46 parser.add_argument('--output_csv', default="data.csv",type=str,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	47 help='Name of the pdf and csv file.')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	48 parser.add_argument('--output_pdf', default="data.pdf",type=str,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	49 help='Name of the pdf and csv file.')
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	50 return parser
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	51
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	52 def compare_read_families(argv):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	53 parser = make_argparser()
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	54 args=parser.parse_args(argv[1:])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	55
6 c4b8222dce29 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 1667d50b521325eea1eada1b16029b1ba1379c5f mheinzl parents: 5 diff changeset	56 firstFile = args.inputFile1
0 9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	57 name1 = args.inputName1
6 c4b8222dce29 planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 1667d50b521325eea1eada1b16029b1ba1379c5f mheinzl parents: 5 diff changeset	58 #firstFile = args.inputName1
5 69f47e0b804e planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 9829d657c07703b117bed2f259ddba432b244312 mheinzl parents: 3 diff changeset	59
0 9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	60 secondFile = args.inputFile2
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	61 name2 = args.inputName2
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	62 thirdFile = args.inputFile3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	63 name3 = args.inputName3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	64 fourthFile = args.inputFile4
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	65 name4 = args.inputName4
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	66
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	67 title_file = args.output_csv
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	68 title_file2 = args.output_pdf
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	69 sep = args.sep
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	70
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	71 if type(sep) is not str or len(sep)>1:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	72 print("Error: --sep must be a single character.")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	73 exit(4)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	74
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	75 plt.rc('figure', figsize=(11.69, 8.27)) # A4 format
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	76 plt.rcParams['patch.edgecolor'] = "black"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	77 plt.rcParams['axes.facecolor'] = "E0E0E0" # grey background color
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	78 plt.rcParams['xtick.labelsize'] = 12
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	79 plt.rcParams['ytick.labelsize'] = 12
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	80
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	81 list_to_plot = []
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	82 label = []
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	83 data_array_list = []
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	84
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	85 with open(title_file, "w") as output_file, PdfPages(title_file2) as pdf:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	86 fig = plt.figure()
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	87 plt.subplots_adjust(bottom=0.25)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	88 if firstFile != str(None):
7 43b16247c35a planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 0641c8994fa786f303f6546d88ce24f15504d9b0 mheinzl parents: 6 diff changeset	89 print(firstFile)
43b16247c35a planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit 0641c8994fa786f303f6546d88ce24f15504d9b0 mheinzl parents: 6 diff changeset	90
0 9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	91 file1 = readFileReferenceFree(firstFile)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	92 integers = numpy.array(file1[:, 0]).astype(int) ## keep original family sizes
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	93
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	94 # for plot: replace all big family sizes by 22
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	95 data1 = numpy.array(file1[:, 0]).astype(int)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	96 bigFamilies = numpy.where(data1 > 20)[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	97 data1[bigFamilies] = 22
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	98
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	99 name1 = name1.split(".tabular")[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	100 list_to_plot.append(data1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	101 label.append(name1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	102 data_array_list.append(file1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	103
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	104 legend = "\n\n\n{}".format(name1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	105 plt.text(0.1, 0.11, legend, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	106 legend1 = "singletons:\nabsolute nr.\n{:,}".format(numpy.bincount(data1)[1])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	107 plt.text(0.4, 0.11, legend1, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	108
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	109 legend3 = "rel. freq\n{:.3f}".format(float(numpy.bincount(data1)[1]) / len(data1))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	110 plt.text(0.5, 0.11, legend3, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	111
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	112 legend4 = "family size > 20:\nabsolute nr.\n{:,}".format(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	113 numpy.bincount(data1)[len(numpy.bincount(data1)) - 1].astype(int))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	114 plt.text(0.6, 0.11, legend4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	115
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	116 legend5 = "rel. freq\n{:.3f}".format(float(numpy.bincount(data1)[len(numpy.bincount(data1)) - 1]) / len(data1))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	117 plt.text(0.7, 0.11, legend5, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	118
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	119 legend6 = "total length\n{:,}".format(len(data1))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	120 plt.text(0.8, 0.11, legend6, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	121
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	122 if secondFile != str(None):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	123 file2 = readFileReferenceFree(secondFile)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	124 data2 = numpy.asarray(file2[:, 0]).astype(int)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	125 bigFamilies2 = numpy.where(data2 > 20)[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	126 data2[bigFamilies2] = 22
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	127
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	128 list_to_plot.append(data2)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	129 name2 = name2.split(".tabular")[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	130 label.append(name2)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	131 data_array_list.append(file2)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	132
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	133 plt.text(0.1, 0.09, name2, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	134
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	135 legend1 = "{:,}".format(numpy.bincount(data2)[1])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	136 plt.text(0.4, 0.09, legend1, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	137
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	138 legend3 = "{:.3f}".format(float(numpy.bincount(data2)[1]) / len(data2))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	139 plt.text(0.5, 0.09, legend3, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	140
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	141 legend4 = "{:,}".format(numpy.bincount(data2)[len(numpy.bincount(data2)) - 1].astype(int))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	142 plt.text(0.6, 0.09, legend4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	143
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	144 legend5 = "{:.3f}".format(float(numpy.bincount(data2)[len(numpy.bincount(data2)) - 1]) / len(data2))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	145 plt.text(0.7, 0.09, legend5, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	146
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	147 legend6 = "{:,}".format(len(data2))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	148 plt.text(0.8, 0.09, legend6, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	149
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	150 if thirdFile != str(None):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	151 file3 = readFileReferenceFree(thirdFile)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	152
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	153 data3 = numpy.asarray(file3[:, 0]).astype(int)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	154 bigFamilies3 = numpy.where(data3 > 20)[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	155 data3[bigFamilies3] = 22
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	156
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	157 list_to_plot.append(data3)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	158 name3 = name3.split(".tabular")[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	159 label.append(name3)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	160 data_array_list.append(file3)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	161
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	162 plt.text(0.1, 0.07, name3, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	163
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	164 legend1 = "{:,}".format(numpy.bincount(data3)[1])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	165 plt.text(0.4, 0.07, legend1, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	166
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	167 legend3 = "{:.3f}".format(float(numpy.bincount(data3)[1]) / len(data3))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	168 plt.text(0.5, 0.07, legend3, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	169
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	170 legend4 = "{:,}".format(numpy.bincount(data3)[len(numpy.bincount(data3)) - 1].astype(int))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	171 plt.text(0.6, 0.07, legend4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	172
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	173 legend5 = "{:.3f}".format(float(numpy.bincount(data3)[len(numpy.bincount(data3)) - 1]) / len(data3))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	174 plt.text(0.7, 0.07, legend5, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	175
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	176 legend6 = "{:,}".format(len(data3))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	177 plt.text(0.8, 0.07, legend6, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	178
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	179 if fourthFile != str(None):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	180 file4 = readFileReferenceFree(fourthFile)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	181
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	182 data4 = numpy.asarray(file4[:, 0]).astype(int)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	183 bigFamilies4 = numpy.where(data4 > 20)[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	184 data4[bigFamilies4] = 22
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	185
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	186 list_to_plot.append(data4)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	187 name4 = name4.split(".tabular")[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	188 label.append(name4)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	189 data_array_list.append(file4)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	190
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	191 plt.text(0.1, 0.05, name4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	192
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	193 legend1 = "{:,}".format(numpy.bincount(data4)[1])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	194 plt.text(0.4, 0.05, legend1, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	195
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	196 legend4 = "{:.3f}".format(float(numpy.bincount(data4)[1]) / len(data4))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	197 plt.text(0.5, 0.05, legend4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	198
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	199 legend4 = "{:,}".format(numpy.bincount(data4)[len(numpy.bincount(data4)) - 1].astype(int))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	200 plt.text(0.6, 0.05, legend4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	201
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	202 legend5 = "{:.3f}".format(float(numpy.bincount(data4)[len(numpy.bincount(data4)) - 1]) / len(data4))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	203 plt.text(0.7, 0.05, legend5, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	204
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	205 legend6 = "{:,}".format(len(data4))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	206 plt.text(0.8, 0.05, legend6, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	207
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	208 maximumX = numpy.amax(numpy.concatenate(list_to_plot))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	209 minimumX = numpy.amin(numpy.concatenate(list_to_plot))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	210
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	211 counts = plt.hist(list_to_plot, bins=range(minimumX, maximumX + 1), stacked=False, edgecolor="black",
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	212 linewidth=1, label=label, align="left", alpha=0.7, rwidth=0.8)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	213
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	214 ticks = numpy.arange(minimumX - 1, maximumX, 1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	215 ticks1 = map(str, ticks)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	216 ticks1[len(ticks1) - 1] = ">20"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	217 plt.xticks(numpy.array(ticks), ticks1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	218
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	219 plt.legend(loc='upper right', fontsize=14, frameon=True, bbox_to_anchor=(0.9, 1))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	220 plt.title("Family Size Distribution", fontsize=14)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	221 plt.xlabel("No. of Family Members", fontsize=14)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	222 plt.ylabel("Absolute Frequency", fontsize=14)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	223 plt.margins(0.01, None)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	224 plt.grid(b=True, which="major", color="#424242", linestyle=":")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	225 pdf.savefig(fig)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	226 plt.close()
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	227
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	228 # write data to CSV file
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	229 output_file.write("Values from family size distribution with all datasets\n")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	230 output_file.write("\nFamily size")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	231 for i in label:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	232 output_file.write("{}{}".format(sep, i))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	233 output_file.write("{}sum".format(sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	234 output_file.write("\n")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	235 j = 0
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	236 for fs in counts[1][0:len(counts[1]) - 1]:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	237 if fs == 21:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	238 fs = ">20"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	239 else:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	240 fs = "={}".format(fs)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	241 output_file.write("FS{}{}".format(fs, sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	242 values_of_fs = []
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	243 if len(label) == 1:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	244 output_file.write("{}{}".format(int(counts[0][j]), sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	245 values_of_fs.append(int(counts[0][j]))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	246 else:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	247 for n in range(len(label)):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	248 output_file.write("{}{}".format(int(counts[0][n][j]), sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	249 values_of_fs.append(int(counts[0][n][j]))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	250 output_file.write("{}\n".format(sum(values_of_fs)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	251 j += 1
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	252 output_file.write("sum{}".format(sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	253 values_for_sum = []
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	254 if len(label) == 1:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	255 output_file.write("{}{}".format(int(sum(counts[0])), sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	256 values_for_sum.append(int(sum(counts[0])))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	257 else:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	258 for i in counts[0]:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	259 output_file.write("{}{}".format(int(sum(i)), sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	260 values_for_sum.append(int(sum(i)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	261
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	262 output_file.write("{}\n".format(sum(values_for_sum)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	263
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	264 ### Family size distribution after DCS and SSCS
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	265 for dataset, data, name_file in zip(list_to_plot, data_array_list, label):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	266 maximumX = numpy.amax(dataset)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	267 minimumX = numpy.amin(dataset)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	268
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	269 tags = numpy.array(data[:, 2])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	270 seq = numpy.array(data[:, 1])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	271 data = numpy.array(dataset)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	272
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	273 # find all unique tags and get the indices for ALL tags, but only once
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	274 u, index_unique, c = numpy.unique(numpy.array(seq), return_counts=True, return_index=True)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	275 d = u[c > 1]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	276
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	277 # get family sizes, tag for duplicates
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	278 duplTags_double = data[numpy.in1d(seq, d)]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	279 duplTags = duplTags_double[0::2] # ab of DCS
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	280 duplTagsBA = duplTags_double[1::2] # ba of DCS
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	281
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	282 duplTags_double_tag = tags[numpy.in1d(seq, d)]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	283 duplTags_double_seq = seq[numpy.in1d(seq, d)]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	284
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	285 # get family sizes for SSCS with no partner
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	286 ab = numpy.where(tags == "ab")[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	287 abSeq = seq[ab]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	288 ab = data[ab]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	289 ba = numpy.where(tags == "ba")[0]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	290 baSeq = seq[ba]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	291 ba = data[ba]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	292
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	293 dataAB = ab[numpy.in1d(abSeq, d, invert=True)]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	294 dataBA = ba[numpy.in1d(baSeq, d, invert=True)]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	295
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	296 # write DCS tags to file
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	297 # with open("DCS information_{}.txt".format(firstFile), "w") as file:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	298 # for t, s, f in zip(duplTags_double_tag, duplTags_double_seq, duplTags_double):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	299 # file.write("{}\t{}\t{}\n".format(t, s, f))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	300
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	301 list1 = [duplTags_double, dataAB, dataBA] # list for plotting
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	302
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	303 ## information for family size >= 3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	304 dataAB_FS3 = dataAB[dataAB >= 3]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	305 dataBA_FS3 = dataBA[dataBA >= 3]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	306 ab_FS3 = ab[ab >= 3]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	307 ba_FS3 = ba[ba >= 3]
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	308
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	309 duplTags_FS3 = duplTags[(duplTags >= 3) & (duplTagsBA >= 3)] # ab+ba with FS>=3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	310 duplTags_FS3_BA = duplTagsBA[(duplTags >= 3) & (duplTagsBA >= 3)] # ba+ab with FS>=3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	311 duplTags_double_FS3 = len(duplTags_FS3)+len(duplTags_FS3_BA) # both ab and ba strands with FS>=3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	312
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	313 fig = plt.figure()
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	314
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	315 plt.subplots_adjust(bottom=0.3)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	316 counts = plt.hist(list1, bins=range(minimumX, maximumX + 1), stacked=True,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	317 label=["duplex", "ab", "ba"], edgecolor="black", linewidth=1,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	318 align="left", color=["#FF0000", "#5FB404", "#FFBF00"])
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	319 # tick labels of x axis
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	320 ticks = numpy.arange(minimumX - 1, maximumX, 1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	321 ticks1 = map(str, ticks)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	322 ticks1[len(ticks1) - 1] = ">20"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	323 plt.xticks(numpy.array(ticks), ticks1)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	324 singl = counts[0][2][0] # singletons
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	325 last = counts[0][2][len(counts[0][0]) - 1] # large families
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	326
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	327 plt.legend(loc='upper right', fontsize=14, bbox_to_anchor=(0.9, 1), frameon=True)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	328 plt.title(name1, fontsize=14)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	329 plt.xlabel("No. of Family Members", fontsize=14)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	330 plt.ylabel("Absolute Frequency", fontsize=14)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	331 plt.margins(0.01, None)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	332 plt.grid(b=True, which="major", color="#424242", linestyle=":")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	333
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	334 ## extra information beneath the plot
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	335 legend = "SSCS ab= \nSSCS ba= \nDCS (total)= \nlength of dataset="
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	336 plt.text(0.1, 0.09, legend, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	337
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	338 legend = "absolute numbers\n\n{:,}\n{:,}\n{:,} ({:,})\n{:,}" \
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	339 .format(len(dataAB), len(dataBA), len(duplTags), len(duplTags_double),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	340 (len(dataAB) + len(dataBA) + len(duplTags)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	341 plt.text(0.35, 0.09, legend, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	342
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	343 legend = "relative frequencies\nunique\n{:.3f}\n{:.3f}\n{:.3f}\n{:,}" \
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	344 .format(float(len(dataAB)) / (len(dataAB) + len(dataBA) + len(duplTags)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	345 float(len(dataBA)) / (len(dataAB) + len(dataBA) + len(duplTags)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	346 float(len(duplTags)) / (len(dataAB) + len(dataBA) + len(duplTags)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	347 (len(dataAB) + len(dataBA) + len(duplTags)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	348 plt.text(0.54, 0.09, legend, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	349
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	350 legend = "total\n{:.3f}\n{:.3f}\n{:.3f} ({:.3f})\n{:,}" \
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	351 .format(float(len(dataAB)) / (len(ab) + len(ba)), float(len(dataBA)) / (len(ab) + len(ba)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	352 float(len(duplTags)) / (len(ab) + len(ba)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	353 float(len(duplTags_double)) / (len(ab) + len(ba)), (len(ab) + len(ba)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	354 plt.text(0.64, 0.09, legend, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	355
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	356 legend1 = "\nsingletons:\nfamily size > 20:"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	357 plt.text(0.1, 0.03, legend1, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	358
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	359 legend4 = "{:,}\n{:,}".format(singl.astype(int), last.astype(int))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	360 plt.text(0.35, 0.03, legend4, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	361
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	362 legend3 = "{:.3f}\n{:.3f}".format(singl / len(data),last / len(data))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	363 plt.text(0.54, 0.03, legend3, size=12, transform=plt.gcf().transFigure)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	364
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	365 pdf.savefig(fig)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	366 plt.close()
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	367
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	368 # write same information to a csv file
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	369 count = numpy.bincount(integers) # original counts of family sizes
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	370 output_file.write("\nDataset:{}{}\n".format(sep, name_file))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	371 output_file.write("max. family size:{}{}\n".format(sep, max(integers)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	372 output_file.write("absolute frequency:{}{}\n".format(sep, count[len(count) - 1]))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	373 output_file.write("relative frequency:{}{:.3f}\n\n".format(sep, float(count[len(count) - 1]) / sum(count)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	374
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	375 output_file.write("{}singletons:{}{}family size > 20:\n".format(sep, sep, sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	376 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	377 "{}absolute nr.{}rel. freq{}absolute nr.{}rel. freq{}total length\n".format(sep, sep, sep, sep, sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	378 output_file.write("{}{}{}{}{:.3f}{}{}{}{:.3f}{}{}\n\n".format(name_file, sep, singl.astype(int), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	379 singl / len(data), sep,last.astype(int), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	380 last / len(data), sep, len(data)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	381
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	382 ## information for FS >= 1
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	383 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	384 "The unique frequencies were calculated from the dataset where the tags occured only once (=ab without DCS, ba without DCS)\n" \
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	385 "Whereas the total frequencies were calculated from the whole dataset (=including the DCS).\n\n")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	386 output_file.write("FS >= 1{}{}unique:{}total:\n".format(sep, sep, sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	387 output_file.write("nr./rel. freq of ab={}{}{}{:.3f}{}{:.3f}\n".format(sep, len(dataAB), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	388 float(len(dataAB)) / (len(dataAB) + len(dataBA) + len( duplTags)), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	389 float(len(dataAB)) / (len(ab) + len(ba))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	390 output_file.write("nr./rel. freq of ba={}{}{}{:.3f}{}{:.3f}\n".format(sep, len(dataBA), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	391 float(len(dataBA)) / (len(dataBA) + len(dataBA) + len(duplTags)), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	392 float(len(dataBA)) / (len(ba) + len(ba))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	393 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	394 "nr./rel. freq of DCS (total)={}{} ({}){}{:.3f}{}{:.3f} ({:.3f})\n".format(sep, len(duplTags), len(duplTags_double), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	395 float(len(duplTags)) / ( len(dataAB) + len( dataBA) + len(duplTags)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	396 sep, float(len(duplTags)) / ( len(ab) + len(ba)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	397 float( len(duplTags_double)) / (len(ab) + len(ba))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	398 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	399 "length of dataset={}{}{}{}{}{}\n".format(sep, (len(dataAB) + len(dataBA) + len(duplTags)), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	400 (len(dataAB) + len(dataBA) + len(duplTags)), sep,(len(ab) + len(ba))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	401 ## information for FS >= 3
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	402 output_file.write("FS >= 3{}{}unique:{}total:\n".format(sep, sep, sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	403 output_file.write("nr./rel. freq of ab={}{}{}{:.3f}{}{:.3f}\n".format(sep, len(dataAB_FS3), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	404 float(len(dataAB_FS3)) / (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	405 sep, float(len(dataAB_FS3)) / ( len(ab_FS3) + len(ba_FS3))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	406 output_file.write("nr./rel. freq of ba={}{}{}{:.3f}{}{:.3f}\n".format(sep, len(dataBA_FS3), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	407 float(len(dataBA_FS3)) / ( len(dataBA_FS3) + len(dataBA_FS3) + len(duplTags_FS3)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	408 sep,float(len(dataBA_FS3)) / (len(ba_FS3) + len(ba_FS3))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	409 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	410 "nr./rel. freq of DCS (total)={}{} ({}){}{:.3f}{}{:.3f} ({:.3f})\n".format(sep, len(duplTags_FS3),duplTags_double_FS3,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	411 sep, float(len( duplTags_FS3)) / (len(dataBA_FS3) + len(duplTags_FS3)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	412 sep, float(len(duplTags_FS3)) / (len(ab_FS3) + len(ba_FS3)),
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	413 float(duplTags_double_FS3) / (len(ab_FS3) + len(ba_FS3))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	414 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	415 "length of dataset={}{}{}{}{}{}\n".format(sep, (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	416 (len(dataAB_FS3) + len(dataBA_FS3) + len(duplTags_FS3)), sep,
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	417 (len(ab_FS3) + len(ba_FS3))))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	418
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	419 output_file.write("\nValues from family size distribution\n")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	420 output_file.write("{}duplex{}ab{}ba{}sum\n".format(sep, sep, sep, sep))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	421 for dx, ab, ba, fs in zip(counts[0][0], counts[0][1], counts[0][2], counts[1]):
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	422 if fs == 21:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	423 fs = ">20"
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	424 else:
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	425 fs = "={}".format(fs)
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	426 ab1 = ab - dx
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	427 ba1 = ba - ab
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	428 output_file.write(
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	429 "FS{}{}{}{}{}{}{}{}{}\n".format(fs, sep, int(dx), sep, int(ab1), sep, int(ba1), sep, int(ba)))
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	430
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	431 print("Files successfully created!")
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	432
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	433 if __name__ == '__main__':
9736b9d04a0b planemo upload for repository https://github.com/monikaheinzl/galaxyProject/tree/master/tools/fsd commit f674213e798956531c935e7b9eb7f444286d0a5e-dirty mheinzl parents: diff changeset	434 sys.exit(compare_read_families(sys.argv))

Mercurial > repos > mheinzl > fsd

annotate fsd.py @ 8:7d5c86d59850 draft