vsnp_add_zero_coverage: vsnp_build

annotate vsnp_build_tables.py @ 0:0ad85e7db2fc draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5"

author	iuc
date	Sun, 27 Sep 2020 10:07:44 +0000
parents
children	aed013f6b13b

rev	line source
0 0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	1 #!/usr/bin/env python
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	2
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	3 import argparse
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	4 import multiprocessing
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	5 import os
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	6 import pandas
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	7 import queue
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	8 import pandas.io.formats.excel
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	9 import re
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	10 from Bio import SeqIO
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	11
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	12 INPUT_JSON_AVG_MQ_DIR = 'input_json_avg_mq_dir'
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	13 INPUT_JSON_DIR = 'input_json_dir'
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	14 INPUT_NEWICK_DIR = 'input_newick_dir'
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	15 # Maximum columns allowed in a LibreOffice
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	16 # spreadsheet is 1024. Excel allows for
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	17 # 16,384 columns, but we'll set the lower
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	18 # number as the maximum. Some browsers
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	19 # (e.g., Firefox on Linux) are configured
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	20 # to use LibreOffice for Excel spreadsheets.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	21 MAXCOLS = 1024
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	22 OUTPUT_EXCEL_DIR = 'output_excel_dir'
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	23
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	24
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	25 def annotate_table(table_df, group, annotation_dict):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	26 for gbk_chrome, pro in list(annotation_dict.items()):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	27 ref_pos = list(table_df)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	28 ref_series = pandas.Series(ref_pos)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	29 ref_df = pandas.DataFrame(ref_series.str.split(':', expand=True).values, columns=['reference', 'position'])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	30 all_ref = ref_df[ref_df['reference'] == gbk_chrome]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	31 positions = all_ref.position.to_frame()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	32 # Create an annotation file.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	33 annotation_file = "%s_annotations.csv" % group
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	34 with open(annotation_file, "a") as fh:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	35 for index, row in positions.iterrows():
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	36 pos = row.position
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	37 try:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	38 aaa = pro.iloc[pro.index.get_loc(int(pos))][['chrom', 'locus', 'product', 'gene']]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	39 try:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	40 chrom, name, locus, tag = aaa.values[0]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	41 print("{}:{}\t{}, {}, {}".format(chrom, pos, locus, tag, name), file=fh)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	42 except ValueError:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	43 # If only one annotation for the entire
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	44 # chromosome (e.g., flu) then having [0] fails
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	45 chrom, name, locus, tag = aaa.values
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	46 print("{}:{}\t{}, {}, {}".format(chrom, pos, locus, tag, name), file=fh)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	47 except KeyError:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	48 print("{}:{}\tNo annotated product".format(gbk_chrome, pos), file=fh)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	49 # Read the annotation file into a data frame.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	50 annotations_df = pandas.read_csv(annotation_file, sep='\t', header=None, names=['index', 'annotations'], index_col='index')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	51 # Remove the annotation_file from disk since both
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	52 # cascade and sort tables are built using the file,
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	53 # and it is opened for writing in append mode.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	54 os.remove(annotation_file)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	55 # Process the data.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	56 table_df_transposed = table_df.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	57 table_df_transposed.index = table_df_transposed.index.rename('index')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	58 table_df_transposed = table_df_transposed.merge(annotations_df, left_index=True, right_index=True)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	59 table_df = table_df_transposed.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	60 return table_df
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	61
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	62
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	63 def excel_formatter(json_file_name, excel_file_name, group, annotation_dict):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	64 pandas.io.formats.excel.header_style = None
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	65 table_df = pandas.read_json(json_file_name, orient='split')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	66 if annotation_dict is not None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	67 table_df = annotate_table(table_df, group, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	68 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	69 table_df = table_df.append(pandas.Series(name='no annotations'))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	70 writer = pandas.ExcelWriter(excel_file_name, engine='xlsxwriter')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	71 table_df.to_excel(writer, sheet_name='Sheet1')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	72 writer_book = writer.book
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	73 ws = writer.sheets['Sheet1']
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	74 format_a = writer_book.add_format({'bg_color': '#58FA82'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	75 format_g = writer_book.add_format({'bg_color': '#F7FE2E'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	76 format_c = writer_book.add_format({'bg_color': '#0000FF'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	77 format_t = writer_book.add_format({'bg_color': '#FF0000'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	78 format_normal = writer_book.add_format({'bg_color': '#FDFEFE'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	79 formatlowqual = writer_book.add_format({'font_color': '#C70039', 'bg_color': '#E2CFDD'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	80 format_ambigous = writer_book.add_format({'font_color': '#C70039', 'bg_color': '#E2CFDD'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	81 format_n = writer_book.add_format({'bg_color': '#E2CFDD'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	82 rows, cols = table_df.shape
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	83 ws.set_column(0, 0, 30)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	84 ws.set_column(1, cols, 2.1)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	85 ws.freeze_panes(2, 1)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	86 format_annotation = writer_book.add_format({'font_color': '#0A028C', 'rotation': '-90', 'align': 'top'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	87 # Set last row.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	88 ws.set_row(rows + 1, cols + 1, format_annotation)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	89 # Make sure that row/column locations don't overlap.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	90 ws.conditional_format(rows - 2, 1, rows - 1, cols, {'type': 'cell', 'criteria': '<', 'value': 55, 'format': formatlowqual})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	91 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'cell', 'criteria': '==', 'value': 'B$2', 'format': format_normal})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	92 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'A', 'format': format_a})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	93 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'G', 'format': format_g})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	94 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'C', 'format': format_c})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	95 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'T', 'format': format_t})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	96 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'S', 'format': format_ambigous})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	97 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'Y', 'format': format_ambigous})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	98 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'R', 'format': format_ambigous})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	99 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'W', 'format': format_ambigous})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	100 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'K', 'format': format_ambigous})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	101 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'M', 'format': format_ambigous})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	102 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'N', 'format': format_n})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	103 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': '-', 'format': format_n})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	104 format_rotation = writer_book.add_format({})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	105 format_rotation.set_rotation(90)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	106 for column_num, column_name in enumerate(list(table_df.columns)):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	107 ws.write(0, column_num + 1, column_name, format_rotation)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	108 format_annotation = writer_book.add_format({'font_color': '#0A028C', 'rotation': '-90', 'align': 'top'})
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	109 # Set last row.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	110 ws.set_row(rows, 400, format_annotation)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	111 writer.save()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	112
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	113
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	114 def get_annotation_dict(gbk_file):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	115 gbk_dict = SeqIO.to_dict(SeqIO.parse(gbk_file, "genbank"))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	116 annotation_dict = {}
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	117 tmp_file = "features.csv"
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	118 # Create a file of chromosomes and features.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	119 for chromosome in list(gbk_dict.keys()):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	120 with open(tmp_file, 'w+') as fh:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	121 for feature in gbk_dict[chromosome].features:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	122 if "CDS" in feature.type or "rRNA" in feature.type:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	123 try:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	124 product = feature.qualifiers['product'][0]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	125 except KeyError:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	126 product = None
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	127 try:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	128 locus = feature.qualifiers['locus_tag'][0]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	129 except KeyError:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	130 locus = None
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	131 try:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	132 gene = feature.qualifiers['gene'][0]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	133 except KeyError:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	134 gene = None
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	135 fh.write("%s\t%d\t%d\t%s\t%s\t%s\n" % (chromosome, int(feature.location.start), int(feature.location.end), locus, product, gene))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	136 # Read the chromosomes and features file into a data frame.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	137 df = pandas.read_csv(tmp_file, sep='\t', names=["chrom", "start", "stop", "locus", "product", "gene"])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	138 # Process the data.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	139 df = df.sort_values(['start', 'gene'], ascending=[True, False])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	140 df = df.drop_duplicates('start')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	141 pro = df.reset_index(drop=True)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	142 pro.index = pandas.IntervalIndex.from_arrays(pro['start'], pro['stop'], closed='both')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	143 annotation_dict[chromosome] = pro
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	144 return annotation_dict
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	145
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	146
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	147 def get_base_file_name(file_path):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	148 base_file_name = os.path.basename(file_path)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	149 if base_file_name.find(".") > 0:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	150 # Eliminate the extension.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	151 return os.path.splitext(base_file_name)[0]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	152 elif base_file_name.find("_") > 0:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	153 # The dot extension was likely changed to
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	154 # the " character.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	155 items = base_file_name.split("_")
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	156 return "_".join(items[0:-1])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	157 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	158 return base_file_name
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	159
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	160
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	161 def output_cascade_table(cascade_order, mqdf, group, annotation_dict):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	162 cascade_order_mq = pandas.concat([cascade_order, mqdf], join='inner')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	163 output_table(cascade_order_mq, "cascade", group, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	164
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	165
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	166 def output_excel(df, type_str, group, annotation_dict, count=None):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	167 # Output the temporary json file that
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	168 # is used by the excel_formatter.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	169 if count is None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	170 if group is None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	171 json_file_name = "%s_order_mq.json" % type_str
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	172 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_table.xlsx" % type_str)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	173 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	174 json_file_name = "%s_%s_order_mq.json" % (group, type_str)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	175 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_%s_table.xlsx" % (group, type_str))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	176 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	177 if group is None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	178 json_file_name = "%s_order_mq_%d.json" % (type_str, count)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	179 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_table_%d.xlsx" % (type_str, count))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	180 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	181 json_file_name = "%s_%s_order_mq_%d.json" % (group, type_str, count)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	182 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_%s_table_%d.xlsx" % (group, type_str, count))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	183 df.to_json(json_file_name, orient='split')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	184 # Output the Excel file.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	185 excel_formatter(json_file_name, excel_file_name, group, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	186
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	187
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	188 def output_sort_table(cascade_order, mqdf, group, annotation_dict):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	189 sort_df = cascade_order.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	190 sort_df['abs_value'] = sort_df.index
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	191 sort_df[['chrom', 'pos']] = sort_df['abs_value'].str.split(':', expand=True)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	192 sort_df = sort_df.drop(['abs_value', 'chrom'], axis=1)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	193 sort_df.pos = sort_df.pos.astype(int)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	194 sort_df = sort_df.sort_values(by=['pos'])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	195 sort_df = sort_df.drop(['pos'], axis=1)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	196 sort_df = sort_df.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	197 sort_order_mq = pandas.concat([sort_df, mqdf], join='inner')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	198 output_table(sort_order_mq, "sort", group, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	199
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	200
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	201 def output_table(df, type_str, group, annotation_dict):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	202 if isinstance(group, str) and group.startswith("dataset"):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	203 # Inputs are single files, not collections,
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	204 # so input file names are not useful for naming
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	205 # output files.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	206 group_str = None
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	207 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	208 group_str = group
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	209 count = 0
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	210 chunk_start = 0
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	211 chunk_end = 0
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	212 column_count = df.shape[1]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	213 if column_count >= MAXCOLS:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	214 # Here the number of columns is greater than
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	215 # the maximum allowed by Excel, so multiple
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	216 # outputs will be produced.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	217 while column_count >= MAXCOLS:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	218 count += 1
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	219 chunk_end += MAXCOLS
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	220 df_of_type = df.iloc[:, chunk_start:chunk_end]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	221 output_excel(df_of_type, type_str, group_str, annotation_dict, count=count)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	222 chunk_start += MAXCOLS
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	223 column_count -= MAXCOLS
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	224 count += 1
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	225 df_of_type = df.iloc[:, chunk_start:]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	226 output_excel(df_of_type, type_str, group_str, annotation_dict, count=count)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	227 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	228 output_excel(df, type_str, group_str, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	229
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	230
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	231 def preprocess_tables(task_queue, annotation_dict, timeout):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	232 while True:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	233 try:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	234 tup = task_queue.get(block=True, timeout=timeout)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	235 except queue.Empty:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	236 break
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	237 newick_file, json_file, json_avg_mq_file = tup
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	238 avg_mq_series = pandas.read_json(json_avg_mq_file, typ='series', orient='split')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	239 # Map quality to dataframe.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	240 mqdf = avg_mq_series.to_frame(name='MQ')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	241 mqdf = mqdf.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	242 # Get the group.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	243 group = get_base_file_name(newick_file)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	244 snps_df = pandas.read_json(json_file, orient='split')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	245 with open(newick_file, 'r') as fh:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	246 for line in fh:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	247 line = re.sub('[:,]', '\n', line)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	248 line = re.sub('[)(]', '', line)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	249 line = re.sub(r'[0-9].\.[0-9].\n', '', line)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	250 line = re.sub('root\n', '', line)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	251 sample_order = line.split('\n')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	252 sample_order = list([_f for _f in sample_order if _f])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	253 sample_order.insert(0, 'root')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	254 tree_order = snps_df.loc[sample_order]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	255 # Count number of SNPs in each column.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	256 snp_per_column = []
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	257 for column_header in tree_order:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	258 count = 0
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	259 column = tree_order[column_header]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	260 for element in column:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	261 if element != column[0]:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	262 count = count + 1
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	263 snp_per_column.append(count)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	264 row1 = pandas.Series(snp_per_column, tree_order.columns, name="snp_per_column")
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	265 # Count number of SNPS from the
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	266 # top of each column in the table.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	267 snp_from_top = []
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	268 for column_header in tree_order:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	269 count = 0
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	270 column = tree_order[column_header]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	271 # for each element in the column
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	272 # skip the first element
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	273 for element in column[1:]:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	274 if element == column[0]:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	275 count = count + 1
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	276 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	277 break
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	278 snp_from_top.append(count)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	279 row2 = pandas.Series(snp_from_top, tree_order.columns, name="snp_from_top")
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	280 tree_order = tree_order.append([row1])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	281 tree_order = tree_order.append([row2])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	282 # In pandas=0.18.1 even this does not work:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	283 # abc = row1.to_frame()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	284 # abc = abc.T --> tree_order.shape (5, 18), abc.shape (1, 18)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	285 # tree_order.append(abc)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	286 # Continue to get error: "*** ValueError: all the input arrays must have same number of dimensions"
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	287 tree_order = tree_order.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	288 tree_order = tree_order.sort_values(['snp_from_top', 'snp_per_column'], ascending=[True, False])
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	289 tree_order = tree_order.T
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	290 # Remove snp_per_column and snp_from_top rows.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	291 cascade_order = tree_order[:-2]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	292 # Output the cascade table.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	293 output_cascade_table(cascade_order, mqdf, group, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	294 # Output the sorted table.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	295 output_sort_table(cascade_order, mqdf, group, annotation_dict)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	296 task_queue.task_done()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	297
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	298
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	299 def set_num_cpus(num_files, processes):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	300 num_cpus = int(multiprocessing.cpu_count())
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	301 if num_files < num_cpus and num_files < processes:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	302 return num_files
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	303 if num_cpus < processes:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	304 half_cpus = int(num_cpus / 2)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	305 if num_files < half_cpus:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	306 return num_files
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	307 return half_cpus
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	308 return processes
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	309
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	310
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	311 if __name__ == '__main__':
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	312 parser = argparse.ArgumentParser()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	313
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	314 parser.add_argument('--input_avg_mq_json', action='store', dest='input_avg_mq_json', required=False, default=None, help='Average MQ json file')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	315 parser.add_argument('--input_newick', action='store', dest='input_newick', required=False, default=None, help='Newick file')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	316 parser.add_argument('--input_snps_json', action='store', dest='input_snps_json', required=False, default=None, help='SNPs json file')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	317 parser.add_argument('--gbk_file', action='store', dest='gbk_file', required=False, default=None, help='Optional gbk file'),
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	318 parser.add_argument('--processes', action='store', dest='processes', type=int, help='User-selected number of processes to use for job splitting')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	319
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	320 args = parser.parse_args()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	321
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	322 if args.gbk_file is not None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	323 # Create the annotation_dict for annotating
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	324 # the Excel tables.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	325 annotation_dict = get_annotation_dict(args.gbk_file)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	326 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	327 annotation_dict = None
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	328
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	329 # The assumption here is that the list of files
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	330 # in both INPUT_NEWICK_DIR and INPUT_JSON_DIR are
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	331 # named such that they are properly matched if
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	332 # the directories contain more than 1 file (i.e.,
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	333 # hopefully the newick file names and json file names
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	334 # will be something like Mbovis-01D6_* so they can be
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	335 # sorted and properly associated with each other).
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	336 if args.input_newick is not None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	337 newick_files = [args.input_newick]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	338 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	339 newick_files = []
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	340 for file_name in sorted(os.listdir(INPUT_NEWICK_DIR)):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	341 file_path = os.path.abspath(os.path.join(INPUT_NEWICK_DIR, file_name))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	342 newick_files.append(file_path)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	343 if args.input_snps_json is not None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	344 json_files = [args.input_snps_json]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	345 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	346 json_files = []
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	347 for file_name in sorted(os.listdir(INPUT_JSON_DIR)):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	348 file_path = os.path.abspath(os.path.join(INPUT_JSON_DIR, file_name))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	349 json_files.append(file_path)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	350 if args.input_avg_mq_json is not None:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	351 json_avg_mq_files = [args.input_avg_mq_json]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	352 else:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	353 json_avg_mq_files = []
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	354 for file_name in sorted(os.listdir(INPUT_JSON_AVG_MQ_DIR)):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	355 file_path = os.path.abspath(os.path.join(INPUT_JSON_AVG_MQ_DIR, file_name))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	356 json_avg_mq_files.append(file_path)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	357
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	358 multiprocessing.set_start_method('spawn')
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	359 queue1 = multiprocessing.JoinableQueue()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	360 queue2 = multiprocessing.JoinableQueue()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	361 num_files = len(newick_files)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	362 cpus = set_num_cpus(num_files, args.processes)
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	363 # Set a timeout for get()s in the queue.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	364 timeout = 0.05
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	365
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	366 for i, newick_file in enumerate(newick_files):
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	367 json_file = json_files[i]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	368 json_avg_mq_file = json_avg_mq_files[i]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	369 queue1.put((newick_file, json_file, json_avg_mq_file))
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	370
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	371 # Complete the preprocess_tables task.
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	372 processes = [multiprocessing.Process(target=preprocess_tables, args=(queue1, annotation_dict, timeout, )) for _ in range(cpus)]
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	373 for p in processes:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	374 p.start()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	375 for p in processes:
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	376 p.join()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	377 queue1.join()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	378
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	379 if queue1.empty():
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	380 queue1.close()
0ad85e7db2fc "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 6a0c9a857c1f4638ef18e106b1f8c0681303acc5" iuc parents: diff changeset	381 queue1.join_thread()

Mercurial > repos > iuc > vsnp_add_zero_coverage

annotate vsnp_build_tables.py @ 0:0ad85e7db2fc draft