annotate vsnp_build_tables.py @ 7:57bd5b859e86 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
author iuc
date Fri, 10 Jun 2022 06:10:23 +0000
parents 532a11cdd818
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
1 #!/usr/bin/env python
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
2
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
3 import argparse
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
4 import itertools
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
5 import multiprocessing
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
6 import os
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
7 import queue
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
8 import re
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
9
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
10 import pandas
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
11 import pandas.io.formats.excel
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
12 from Bio import SeqIO
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
13
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
14 # Maximum columns allowed in a LibreOffice
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
15 # spreadsheet is 1024. Excel allows for
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
16 # 16,384 columns, but we'll set the lower
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
17 # number as the maximum. Some browsers
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
18 # (e.g., Firefox on Linux) are configured
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
19 # to use LibreOffice for Excel spreadsheets.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
20 MAXCOLS = 1024
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
21 OUTPUT_EXCEL_DIR = 'output_excel_dir'
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
22 INPUT_JSON_AVG_MQ_DIR = 'input_json_avg_mq_dir'
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
23 INPUT_JSON_DIR = 'input_json_dir'
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
24 INPUT_NEWICK_DIR = 'input_newick_dir'
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
25
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
26
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
27 def annotate_table(table_df, group, annotation_dict):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
28 for gbk_chrome, pro in list(annotation_dict.items()):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
29 ref_pos = list(table_df)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
30 ref_series = pandas.Series(ref_pos)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
31 ref_df = pandas.DataFrame(ref_series.str.split(':', expand=True).values, columns=['reference', 'position'])
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
32 all_ref = ref_df[ref_df['reference'] == gbk_chrome]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
33 positions = all_ref.position.to_frame()
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
34 # Create an annotation file.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
35 annotation_file = "%s_annotations.csv" % group
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
36 with open(annotation_file, "a") as fh:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
37 for _, row in positions.iterrows():
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
38 pos = row.position
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
39 try:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
40 aaa = pro.iloc[pro.index.get_loc(int(pos))][['chrom', 'locus', 'product', 'gene']]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
41 try:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
42 chrom, name, locus, tag = aaa.values[0]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
43 print("{}:{}\t{}, {}, {}".format(chrom, pos, locus, tag, name), file=fh)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
44 except ValueError:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
45 # If only one annotation for the entire
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
46 # chromosome (e.g., flu) then having [0] fails
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
47 chrom, name, locus, tag = aaa.values
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
48 print("{}:{}\t{}, {}, {}".format(chrom, pos, locus, tag, name), file=fh)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
49 except KeyError:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
50 print("{}:{}\tNo annotated product".format(gbk_chrome, pos), file=fh)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
51 # Read the annotation file into a data frame.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
52 annotations_df = pandas.read_csv(annotation_file, sep='\t', header=None, names=['index', 'annotations'], index_col='index')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
53 # Remove the annotation_file from disk since both
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
54 # cascade and sort tables are built using the file,
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
55 # and it is opened for writing in append mode.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
56 os.remove(annotation_file)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
57 # Process the data.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
58 table_df_transposed = table_df.T
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
59 table_df_transposed.index = table_df_transposed.index.rename('index')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
60 table_df_transposed = table_df_transposed.merge(annotations_df, left_index=True, right_index=True)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
61 table_df = table_df_transposed.T
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
62 return table_df
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
63
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
64
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
65 def excel_formatter(json_file_name, excel_file_name, group, annotation_dict):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
66 pandas.io.formats.excel.header_style = None
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
67 table_df = pandas.read_json(json_file_name, orient='split')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
68 if annotation_dict is not None:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
69 table_df = annotate_table(table_df, group, annotation_dict)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
70 else:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
71 table_df = table_df.append(pandas.Series(name='no annotations'))
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
72 writer = pandas.ExcelWriter(excel_file_name, engine='xlsxwriter')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
73 table_df.to_excel(writer, sheet_name='Sheet1')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
74 writer_book = writer.book
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
75 ws = writer.sheets['Sheet1']
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
76 format_a = writer_book.add_format({'bg_color': '#58FA82'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
77 format_g = writer_book.add_format({'bg_color': '#F7FE2E'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
78 format_c = writer_book.add_format({'bg_color': '#0000FF'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
79 format_t = writer_book.add_format({'bg_color': '#FF0000'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
80 format_normal = writer_book.add_format({'bg_color': '#FDFEFE'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
81 formatlowqual = writer_book.add_format({'font_color': '#C70039', 'bg_color': '#E2CFDD'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
82 format_ambigous = writer_book.add_format({'font_color': '#C70039', 'bg_color': '#E2CFDD'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
83 format_n = writer_book.add_format({'bg_color': '#E2CFDD'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
84 rows, cols = table_df.shape
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
85 ws.set_column(0, 0, 30)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
86 ws.set_column(1, cols, 2.1)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
87 ws.freeze_panes(2, 1)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
88 format_annotation = writer_book.add_format({'font_color': '#0A028C', 'rotation': '-90', 'align': 'top'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
89 # Set last row.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
90 ws.set_row(rows + 1, cols + 1, format_annotation)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
91 # Make sure that row/column locations don't overlap.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
92 ws.conditional_format(rows - 2, 1, rows - 1, cols, {'type': 'cell', 'criteria': '<', 'value': 55, 'format': formatlowqual})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
93 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'cell', 'criteria': '==', 'value': 'B$2', 'format': format_normal})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
94 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'A', 'format': format_a})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
95 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'G', 'format': format_g})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
96 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'C', 'format': format_c})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
97 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'T', 'format': format_t})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
98 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'S', 'format': format_ambigous})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
99 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'Y', 'format': format_ambigous})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
100 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'R', 'format': format_ambigous})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
101 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'W', 'format': format_ambigous})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
102 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'K', 'format': format_ambigous})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
103 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'M', 'format': format_ambigous})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
104 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': 'N', 'format': format_n})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
105 ws.conditional_format(2, 1, rows - 2, cols, {'type': 'text', 'criteria': 'containing', 'value': '-', 'format': format_n})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
106 format_rotation = writer_book.add_format({})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
107 format_rotation.set_rotation(90)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
108 for column_num, column_name in enumerate(list(table_df.columns)):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
109 ws.write(0, column_num + 1, column_name, format_rotation)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
110 format_annotation = writer_book.add_format({'font_color': '#0A028C', 'rotation': '-90', 'align': 'top'})
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
111 # Set last row.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
112 ws.set_row(rows, 400, format_annotation)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
113 writer.save()
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
114
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
115
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
116 def get_annotation_dict(gbk_file):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
117 gbk_dict = SeqIO.to_dict(SeqIO.parse(gbk_file, "genbank"))
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
118 annotation_dict = {}
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
119 tmp_file = "features.csv"
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
120 # Create a file of chromosomes and features.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
121 for chromosome in list(gbk_dict.keys()):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
122 with open(tmp_file, 'w+') as fh:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
123 for feature in gbk_dict[chromosome].features:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
124 if "CDS" in feature.type or "rRNA" in feature.type:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
125 try:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
126 product = feature.qualifiers['product'][0]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
127 except KeyError:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
128 product = None
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
129 try:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
130 locus = feature.qualifiers['locus_tag'][0]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
131 except KeyError:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
132 locus = None
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
133 try:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
134 gene = feature.qualifiers['gene'][0]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
135 except KeyError:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
136 gene = None
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
137 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))
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
138 # Read the chromosomes and features file into a data frame.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
139 df = pandas.read_csv(tmp_file, sep='\t', names=["chrom", "start", "stop", "locus", "product", "gene"])
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
140 # Process the data.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
141 df = df.sort_values(['start', 'gene'], ascending=[True, False])
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
142 df = df.drop_duplicates('start')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
143 pro = df.reset_index(drop=True)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
144 pro.index = pandas.IntervalIndex.from_arrays(pro['start'], pro['stop'], closed='both')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
145 annotation_dict[chromosome] = pro
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
146 return annotation_dict
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
147
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
148
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
149 def get_sample_name(file_path):
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
150 base_file_name = os.path.basename(file_path)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
151 if base_file_name.find(".") > 0:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
152 # Eliminate the extension.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
153 return os.path.splitext(base_file_name)[0]
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
154 return base_file_name
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
155
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
156
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
157 def output_excel(df, type_str, group, annotation_dict, count=None):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
158 # Output the temporary json file that
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
159 # is used by the excel_formatter.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
160 if count is None:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
161 if group is None:
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
162 json_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_order_mq.json" % type_str)
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
163 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_table.xlsx" % type_str)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
164 else:
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
165 json_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_%s_order_mq.json" % (group, type_str))
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
166 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_%s_table.xlsx" % (group, type_str))
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
167 else:
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
168 # The table has more columns than is allowed by the
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
169 # MAXCOLS setting, so multiple files will be produced
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
170 # as an output collection.
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
171 if group is None:
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
172 json_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_order_mq_%d.json" % (type_str, count))
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
173 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_table_%d.xlsx" % (type_str, count))
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
174 else:
1
b03e88e7bb1d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2e312886647244b416c64eca91e1a61dd1be939b"
iuc
parents: 0
diff changeset
175 json_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_%s_order_mq_%d.json" % (group, type_str, count))
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
176 excel_file_name = os.path.join(OUTPUT_EXCEL_DIR, "%s_%s_table_%d.xlsx" % (group, type_str, count))
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
177 df.to_json(json_file_name, orient='split')
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
178 # Output the Excel file.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
179 excel_formatter(json_file_name, excel_file_name, group, annotation_dict)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
180
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
181
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
182 def output_table(df, type_str, group, annotation_dict):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
183 if isinstance(group, str) and group.startswith("dataset"):
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
184 # Inputs are single files, not collections,
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
185 # so input file names are not useful for naming
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
186 # output files.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
187 group_str = None
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
188 else:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
189 group_str = group
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
190 count = 0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
191 chunk_start = 0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
192 chunk_end = 0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
193 column_count = df.shape[1]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
194 if column_count >= MAXCOLS:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
195 # Here the number of columns is greater than
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
196 # the maximum allowed by Excel, so multiple
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
197 # outputs will be produced.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
198 while column_count >= MAXCOLS:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
199 count += 1
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
200 chunk_end += MAXCOLS
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
201 df_of_type = df.iloc[:, chunk_start:chunk_end]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
202 output_excel(df_of_type, type_str, group_str, annotation_dict, count=count)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
203 chunk_start += MAXCOLS
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
204 column_count -= MAXCOLS
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
205 count += 1
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
206 df_of_type = df.iloc[:, chunk_start:]
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
207 output_excel(df_of_type, type_str, group_str, annotation_dict, count=count)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
208 else:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
209 output_excel(df, type_str, group_str, annotation_dict)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
210
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
211
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
212 def preprocess_tables(task_queue, annotation_dict, timeout):
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
213 while True:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
214 try:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
215 tup = task_queue.get(block=True, timeout=timeout)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
216 except queue.Empty:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
217 break
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
218 newick_file, json_file, json_avg_mq_file = tup
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
219 # Get the group.
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
220 group = get_sample_name(newick_file)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
221 snps_df = pandas.read_json(json_file, orient='split')
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
222 with open(newick_file, 'r') as fh:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
223 for line in fh:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
224 line = re.sub('[:,]', '\n', line)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
225 line = re.sub('[)(]', '', line)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
226 line = re.sub(r'[0-9].*\.[0-9].*\n', '', line)
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
227 line = re.sub("'", '', line)
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
228 line = re.sub('root\n', '', line)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
229 sample_order = line.split('\n')
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
230 sample_order = list(filter(None, sample_order))
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
231 sample_order.insert(0, 'root')
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
232 tree_order_df = snps_df.loc[sample_order]
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
233 # Output the sorted table.
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
234 output_table(tree_order_df, "sort", group, annotation_dict)
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
235 # Count number of SNPs in each column.
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
236 snp_per_column = []
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
237 for column_header in tree_order_df:
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
238 count = 0
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
239 column = tree_order_df[column_header]
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
240 for element in column:
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
241 # column[0] is top row/root/reference,
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
242 # element is everything below it.
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
243 if element != column[0] and element != '-':
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
244 count = count + 1
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
245 snp_per_column.append(count)
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
246 row1 = pandas.Series(snp_per_column, tree_order_df.columns, name="snp_per_column")
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
247 # Count number of SNPS from the
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
248 # top of each column in the table.
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
249 snp_from_top = []
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
250 for column_header in tree_order_df:
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
251 count = 0
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
252 column = tree_order_df[column_header]
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
253 index_list_of_ref_differences = []
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
254 for ind, list_item in enumerate(column[1:].to_list()):
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
255 if list_item not in [column[0], '-']:
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
256 index_list_of_ref_differences.append(ind)
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
257 c = itertools.count()
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
258 val = max((list(g) for _, g in itertools.groupby(index_list_of_ref_differences, lambda x: x - next(c))), key=len)
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
259 # Starting row number with longest continous SNPs in column
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
260 snp_from_top.append(val[0])
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
261 row2 = pandas.Series(snp_from_top, tree_order_df.columns, name="snp_from_top")
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
262 tree_order_df = tree_order_df.append([row1])
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
263 tree_order_df = tree_order_df.append([row2])
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
264 tree_order_df = tree_order_df.T
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
265 tree_order_df = tree_order_df.sort_values(['snp_from_top', 'snp_per_column'], ascending=[True, False])
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
266 tree_order_df = tree_order_df.T
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
267 # Remove snp_per_column and snp_from_top rows.
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
268 cascade_order_df = tree_order_df[:-2]
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
269 # Output the cascade table.
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
270 output_table(cascade_order_df, "cascade", group, annotation_dict)
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
271 task_queue.task_done()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
272
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
273
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
274 if __name__ == '__main__':
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
275 parser = argparse.ArgumentParser()
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
276
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
277 parser.add_argument('--input_avg_mq_json', action='store', dest='input_avg_mq_json', required=False, default=None, help='Average MQ json file')
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
278 parser.add_argument('--input_newick', action='store', dest='input_newick', required=False, default=None, help='Newick file')
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
279 parser.add_argument('--input_snps_json', action='store', dest='input_snps_json', required=False, default=None, help='SNPs json file')
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
280 parser.add_argument('--gbk_file', action='store', dest='gbk_file', required=False, default=None, help='Optional gbk file'),
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
281 parser.add_argument('--processes', action='store', dest='processes', type=int, help='User-selected number of processes to use for job splitting')
0
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
282
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
283 args = parser.parse_args()
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
284
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
285 if args.gbk_file is not None:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
286 # Create the annotation_dict for annotating
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
287 # the Excel tables.
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
288 annotation_dict = get_annotation_dict(args.gbk_file)
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
289 else:
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
290 annotation_dict = None
12f2b14549f6 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 524a39e08f2bea8b8754284df606ff8dd27ed24b"
iuc
parents:
diff changeset
291
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
292 # The assumption here is that the list of files
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
293 # in both INPUT_NEWICK_DIR and INPUT_JSON_DIR are
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
294 # named such that they are properly matched if
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
295 # the directories contain more than 1 file (i.e.,
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
296 # hopefully the newick file names and json file names
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
297 # will be something like Mbovis-01D6_* so they can be
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
298 # sorted and properly associated with each other).
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
299 if args.input_newick is not None:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
300 newick_files = [args.input_newick]
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
301 else:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
302 newick_files = []
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
303 for file_name in sorted(os.listdir(INPUT_NEWICK_DIR)):
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
304 file_path = os.path.abspath(os.path.join(INPUT_NEWICK_DIR, file_name))
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
305 newick_files.append(file_path)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
306 if args.input_snps_json is not None:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
307 json_files = [args.input_snps_json]
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
308 else:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
309 json_files = []
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
310 for file_name in sorted(os.listdir(INPUT_JSON_DIR)):
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
311 file_path = os.path.abspath(os.path.join(INPUT_JSON_DIR, file_name))
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
312 json_files.append(file_path)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
313 if args.input_avg_mq_json is not None:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
314 json_avg_mq_files = [args.input_avg_mq_json]
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
315 else:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
316 json_avg_mq_files = []
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
317 for file_name in sorted(os.listdir(INPUT_JSON_AVG_MQ_DIR)):
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
318 file_path = os.path.abspath(os.path.join(INPUT_JSON_AVG_MQ_DIR, file_name))
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
319 json_avg_mq_files.append(file_path)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
320
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
321 multiprocessing.set_start_method('spawn')
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
322 queue1 = multiprocessing.JoinableQueue()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
323 queue2 = multiprocessing.JoinableQueue()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
324 num_files = len(newick_files)
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
325 # Set a timeout for get()s in the queue.
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
326 timeout = 0.05
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
327
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
328 for i, newick_file in enumerate(newick_files):
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
329 json_file = json_files[i]
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
330 json_avg_mq_file = json_avg_mq_files[i]
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
331 queue1.put((newick_file, json_file, json_avg_mq_file))
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
332
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
333 # Complete the preprocess_tables task.
7
57bd5b859e86 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit c38fd63f7980c70390d104a73ba4c72b266444c3
iuc
parents: 6
diff changeset
334 processes = [multiprocessing.Process(target=preprocess_tables, args=(queue1, annotation_dict, timeout, )) for _ in range(args.processes)]
5
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
335 for p in processes:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
336 p.start()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
337 for p in processes:
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
338 p.join()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
339 queue1.join()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
340
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
341 if queue1.empty():
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
342 queue1.close()
a8560decb495 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/vsnp commit 2a94c64d6c7236550bf483d2ffc4e86248c63aab"
iuc
parents: 1
diff changeset
343 queue1.join_thread()