Mercurial > repos > greg > validate_affy_metadata

comparison validate_affy_metadata.py @ 0:80d672b3e6dd draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author greg
date Thu, 15 Aug 2019 13:17:41 -0400
parents
children b5f5c3d0f349
comparison
equal deleted inserted replaced
-1:000000000000 0:80d672b3e6dd
1 #!/usr/bin/env python
2 """
3 Validate the metadata file associated with Affymetrix 96 well plate data.
4 """
5 import argparse
6 import datetime
7 import decimal
8 import re
9 import shutil
10 import sys
11
12 parser = argparse.ArgumentParser()
13 parser.add_argument('--input', dest='input', help='Metadata file for Affymetrix 96 well plate data')
14 parser.add_argument('--output', dest='output', help='Output dataset'),
15 args = parser.parse_args()
16
17 EMAIL_MAX_LEN = 255
18 VALID_EMAIL_RE = re.compile("[^@]+@[^@]+\.[^@]+")
19
20
21 def add_error_msg(accumulated_msgs, msg):
22 return "%s\n%s" % (accumulated_msgs, msg)
23
24
25 def empty_value(line_no, label, accumulated_msgs):
26 return add_error_msg(accumulated_msgs, "The required %s value is missing on line %d." % (label, line_no))
27
28
29 def stop_error(msg):
30 sys.exit(msg)
31
32
33 def string_as_boolean_string(string):
34 if str(string).lower() in ['true', 'yes', 'on', '1']:
35 return 'True'
36 else:
37 return 'False'
38
39
40 def validate_date_string(line_no, date_string, column, accumulated_msgs):
41 if len(date_string) == 0:
42 return accumulated_msgs
43 try:
44 datetime.datetime.strptime(date_string, '%Y-%m-%d')
45 return accumulated_msgs
46 except ValueError:
47 return add_error_msg(accumulated_msgs, "Line %d contains an incorrect date format (%s must be YYYY-MM-DD) for column %s." % (line_no, date_string, column))
48
49
50 def validate_decimal(line_no, decimal_string, column, accumulated_msgs):
51 try:
52 decimal.Decimal(decimal_string)
53 return accumulated_msgs
54 except Exception:
55 return add_error_msg(accumulated_msgs, "Line %d contains an incorrect decimal value (%s) for column %s." % (line_no, decimal_string, column))
56
57
58 def validate_email(line_no, email, accumulated_msgs):
59 if not (VALID_EMAIL_RE.match(email)):
60 return add_error_msg(accumulated_msgs, "Line %d contains an invalid email address (%s). " % (line_no, email))
61 elif len(email) > EMAIL_MAX_LEN:
62 return add_error_msg(accumulated_msgs, "Line %d contains an email address (%) that is longer than the maximum length, %d characters." % (line_no, email))
63 return accumulated_msgs
64
65
66 accumulated_msgs = ""
67 # Parse the input file, skipping the header, and validating
68 # that each data line consists of 31 comma-separated items.
69 with open(args.input, "r") as ih:
70 for i, line in enumerate(ih):
71 if i == 0:
72 # Skip the header.
73 continue
74 # Keep 1-based line value for error messages.
75 line_no = i + 1
76 line = line.rstrip("\r\n")
77 if i > 97:
78 accumulated_msgs = add_error_msg(accumulated_msgs, "The input file contains more than 97 lines (must be 1 header line and no more than 96 data lines).")
79 stop_error(accumulated_msgs)
80 items = line.split("\t")
81 if len(items) != 32:
82 accumulated_msgs = add_error_msg(accumulated_msgs, "Line %d contains %s columns, (must be 32)." % (line_no, len(items)))
83 stop_error(accumulated_msgs)
84 # Required and validated.
85 # Required.
86 user_specimen_id = items[0]
87 if len(user_specimen_id) == 0:
88 accumulated_msgs = empty_value(line_no, "user_specimen_id", accumulated_msgs)
89 # Optional.
90 field_call = items[1]
91 # Optional.
92 bcoral_genet_id = items[2]
93 # Optional.
94 bsym_genet_id = items[3]
95 # Required.
96 reef = items[4]
97 if len(reef) == 0:
98 accumulated_msgs = empty_value(line_no, "reef", accumulated_msgs)
99 # Required.
100 region = items[5]
101 if len(region) == 0:
102 accumulated_msgs = empty_value(line_no, "region", accumulated_msgs)
103 # Required and validated.
104 latitude = items[6]
105 accumulated_msgs = validate_decimal(line_no, latitude, "latitude", accumulated_msgs)
106 # Required and validated.
107 longitude = items[7]
108 accumulated_msgs = validate_decimal(line_no, longitude, "longitude", accumulated_msgs)
109 # Optional.
110 geographic_origin = items[8]
111 # Optional.
112 colony_location = items[9]
113 # Optional.
114 depth = items[10]
115 # Optional.
116 disease_resist = items[11]
117 # Optional.
118 bleach_resist = items[12]
119 # Optional.
120 mortality = items[13]
121 # Optional.
122 tle = items[14]
123 # Optional.
124 spawning = string_as_boolean_string(items[15])
125 # Required.
126 collector_last_name = items[16]
127 if len(collector_last_name) == 0:
128 accumulated_msgs = empty_value(line_no, "collector_last_name", accumulated_msgs)
129 # Required.
130 collector_first_name = items[17]
131 if len(collector_first_name) == 0:
132 accumulated_msgs = empty_value(line_no, "collector_first_name", accumulated_msgs)
133 # Required.
134 org = items[18]
135 if len(org) == 0:
136 accumulated_msgs = empty_value(line_no, "org", accumulated_msgs)
137 # Required and validated.
138 collection_date = items[19]
139 accumulated_msgs = validate_date_string(line_no, collection_date, "collection_date", accumulated_msgs)
140 # Required and validated.
141 contact_email = items[20]
142 accumulated_msgs = validate_email(line_no, contact_email, accumulated_msgs)
143 # Required.
144 seq_facility = items[21]
145 if len(seq_facility) == 0:
146 accumulated_msgs = empty_value(line_no, "seq_facility", accumulated_msgs)
147 # Optional.
148 array_version = items[22]
149 # Optional.
150 public = string_as_boolean_string(items[23])
151 # Optional.
152 public_after_date = items[24]
153 accumulated_msga = validate_date_string(line_no, public_after_date, "public_after_date", accumulated_msgs)
154 # Required and validated.
155 sperm_motility = items[25]
156 accumulated_msgs = validate_decimal(line_no, sperm_motility, "sperm_motility", accumulated_msgs)
157 # Required and validated.
158 healing_time = items[26]
159 accumulated_msgs = validate_decimal(line_no, healing_time, "healing_time", accumulated_msgs)
160 # Optional.
161 dna_extraction_method = items[27]
162 # Optional.
163 dna_concentration = items[28]
164 # If dna_concentration has a value, then it must be decimal.
165 if len(dna_concentration) > 0:
166 accumulated_msgs = validate_decimal(line_no, dna_concentration, "dna_concentration", accumulated_msgs)
167 # Optional.
168 registry_id = items[29]
169 # Optional.
170 result_folder_name = items[30]
171 # Optional.
172 plate_barcode = items[31]
173
174
175 if len(accumulated_msgs) > 0:
176 stop_error(accumulated_msgs)
177
178 shutil.copyfile(args.input, args.output)