validate_gff3.pl

   1 #!/usr/bin/perl
   2
   3 =head1 NAME
   4
   5 validate_gff3.pl
   6
   7 =head1 SYNOPSIS
   8
   9  validate_gff3.pl -gff3_file <gff3_file> [-ontology_file <ontology_file1> -ontology_file <ontology_file1> ...]
  10                   -out <out_file_prefix> -config <config_file>
  11                   [-db_type <db_type>] [-db_name <db_name>] [-username <username>] [-password <password>]
  12                   [-verbose <0|1|2>] [-silent <0|1>]
  13
  14 =head1 DESCRIPTION
  15
  16 This script analyzes a gff3 file and validates a number of points. It uses the GFF3::Validator module for analysis. For
  17 further information on analysis steps, please refer to validate_gff3.pod.
  18
  19 =head1 USAGE
  20
  21 The script uses a  MySQL or SQLite database to analyze the gff3 file. The gff3 file is parsed and
  22 content relevant to the analysis is loaded into the database. Use of database (as opposed to
  23 performing analysis in memory) makes processing of large files feasible and significatly
  24 increases overall processing speed. At the end of the analysis a
  25 report is generated that lists errors and warnings ordered by line numbers. The report file
  26 can be easily processed using grep and other Unix text processing tools.
  27
  28 The usage of the script follows with descriptions of command-line parameters:
  29
  30  validate_gff3.pl -gff3_file <gff3_file> [-ontology_file <ontology_file1> -ontology_file <ontology_file1> ...]
  31                   -out <out_file_prefix> -config <config_file>
  32                   [-db_type <db_type>] [-db_dir <db_dir>]
  33                   [-dbname <dbname>] [-username <username>] [-password <password>]
  34                   [-verbose <0|1|2>] [-silent <0|1>]
  35
  36  -gff3_file     : (Required) Name of gff3 file to process.
  37  -ontology_file : (Optional) Name of ontology file, multiple files can be specified.
  38                   Command-line ontology files and ontology files provided as directives
  39                   are merged and used for analysis. If neither is provided or is not accesible,
  40                   default ontology file is retrieved and used.
  41  -out           : (Required) Prefix to name log and report files, these become <out>.log and <out>.report
  42  -config        : (Required) Name of config file (see documentation in validate_gff3.cfg provided in the package
  43                   for further details).
  44  -db_type       : (Optional) Type of database ('mysql' or 'sqlite').
  45                   Defaults to 'mysql'.
  46  -db_dir        : (Optional) Directory to store temp sqlite database files
  47                   If not available, retrieved from config file (temp_dir param)
  48  -dbname        : (Optional) Name of MySQL database/SQLite db file to use for analysis.
  49                   If not available, retrieved from config file.
  50                   If db_type is 'sqlite' and no dbname is specified and none available in config file, a temp db is used
  51  -username      : (Optional) Username for analysis database (must have write privileges).
  52                   If not available, retrieved from config file.
  53                   If not available, defaults to "".
  54  -password      : (Optional) Password for analysis database.
  55                   If not available, retrieved from config file.
  56                   If not available, defaults to "".
  57  -verbose       : (Optional) Verbosity of logging.
  58                   Values:
  59                   1: Initialization information
  60                   2: + Progress information
  61                   3: + Error messages
  62                   If not available, defaults to 2.
  63  -silent        : (Optional) Whether to suppress logging to screen
  64                   Values:
  65                   0: Log to screen
  66                   1: Don't log to screen
  67                   If not available, defaults to 0.
  68  -max_messages  : (Optional) Whether to report all errors/warnings
  69                   Values:
  70                   0: Report all messages
  71                   <number>: Exit and report after <number> messages
  72                   If not available, defaults to 0.
  73
  74 =cut
  75
  76 use strict;
  77
  78 use FindBin;
  79 use File::Spec;
  80 use File::Temp;
  81 use lib "$FindBin::RealBin/lib";
  82
  83 use GFF3::Validator;
  84 use Carp;
  85 use Getopt::Long;
  86
  87 # Usage
  88 my $usage = qq[$FindBin::Script -gff3_file <gff3_file> [-ontology_file <ontology_file1> -ontology_file <ontology_file1> ...]
  89                   -out <out_file_prefix>
  90                   [-config <config_file>]
  91                   [-db_type <db_type>] [-db_dir <db_dir>]
  92                   [-dbname <dbname>] [-username <username>] [-password <password>]
  93                   [-verbose <0|1|2>] [-silent <0|1>]];
  94
  95 # Parse command-line params
  96 my $gff3_file;
  97 my @ontology_files;
  98 my $out;
  99 my $config;
 100 my $db_type;
 101 my $db_dir;
 102 my $dbname;
 103 my $username;
 104 my $password;
 105 my $verbose;
 106 my $silent;
 107 my $max_messages;
 108
 109 my $result = GetOptions ("gff3_file=s"     => \$gff3_file,
 110                          "ontology_files=s" => \@ontology_files,
 111                          "out=s"           => \$out,
 112                          "config=s"        => \$config,
 113                          "db_type=s"       => \$db_type,
 114                          "db_dir=s"        => \$db_dir,
 115                          "dbname=s"        => \$dbname,
 116                          "username=s"      => \$username,
 117                          "password=s"      => \$password,
 118                          "verbose=s"       => \$verbose,
 119                          "silent=s"        => \$silent,
 120                          "max_messages=s"  => \$max_messages,
 121                          ) or die("Usage: $usage\n");
 122
 123 # Check command-line params
 124 if (!$gff3_file or !$out ) {
 125     die("Usage: $usage\n");
 126     }
 127
 128 # Parse config file
 129 $config ||= "$FindBin::RealBin/validate_gff3.cfg";
 130 my $config_obj = Config::General->new(-ConfigFile => $config, -CComments => 0);
 131 my %config = $config_obj->getall;
 132
 133 # Populate defaults from config
 134 $db_type = lc($db_type) || 'mysql';
 135 croak("Unrecognized database type ($db_type)!") unless $db_type =~ /^(mysql|sqlite)$/;
 136
 137 $db_dir ||= $config{temp_dir};
 138 unless( -d $db_dir ) {
 139   mkdir $db_dir or die "$db_dir does not exist, and can't create it\n";
 140   chmod 0777, $db_dir or warn "WARNING: could not set global temp dir $db_dir world-writable\n";
 141 }
 142 croak("Cannot determine db dir!") unless $db_dir;
 143
 144 my $datasource;
 145 if ($dbname && $db_type eq 'mysql') {
 146     $datasource = "DBI:mysql:dbname=$dbname";
 147     }
 148 elsif ($dbname && $db_type eq 'sqlite') {
 149     $datasource = "DBI:SQLite:dbname=$dbname";
 150     }
 151 elsif (!$dbname && $db_type eq 'mysql') {
 152     $datasource = $config{datasource};
 153     }
 154 elsif (!$dbname && $db_type eq 'sqlite') {
 155     my ($temp_fh, $temp_file) = File::Temp::tempfile("validate_gff3_sqlite_XXXXX",
 156                                                      DIR     => $db_dir,
 157                                                      SUFFIX  => '.db',
 158                                                      UNLINK  => 1);
 159
 160     $datasource = "DBI:SQLite:dbname=$temp_file";
 161     }
 162 else {
 163     $datasource = $config{datasource}; # Placeholder
 164     }
 165 croak("Cannot determine database name!") unless $datasource;
 166
 167 # Prepare params
 168 my $log_file = "$out.log";
 169 my $report_file = "$out.report";
 170
 171 # Create validator object
 172 my $validator = GFF3::Validator->new(-config         => $config,
 173                                      -gff3_file      => $gff3_file,
 174                                      -datasource     => $datasource,
 175                                      -username       => $username,
 176                                      -password       => $password,
 177                                      -verbose        => $verbose,
 178                                      -silent         => $silent,
 179                                      -max_messages   => $max_messages,
 180                                      -log_file       => $log_file,
 181                                      -report_file    => $report_file,
 182                                      -ontology_files => \@ontology_files,
 183                                      -table_id       => "", # Currently do not use table id feature within the command-line version
 184                                      );
 185
 186 # Create/Reset tables to store the data
 187 $validator->create_tables;
 188
 189 # Load gff3 analysis database
 190 $validator->load_analysis_dbs;
 191
 192 # Validate unique ids
 193 $validator->validate_unique_ids;
 194
 195 # Load ontology(s) into memory
 196 $validator->load_ontology;
 197
 198 # Validate ontology terms
 199 $validator->validate_ontology_terms;
 200
 201 # Validate parentage
 202 $validator->validate_parentage;
 203
 204 # Validate derives_from
 205 $validator->validate_derives_from;
 206
 207 # Dump an error report
 208 $validator->dump_report;
 209
 210 # Cleanup
 211 # $validator->cleanup; # Currently, do not clean up within the command-line version
 212
 213 $validator->log("# [END]");
 214
 215 =head1 SEE ALSO
 216
 217 =head1 AUTHOR
 218
 219 Payan Canaran <canaran@cshl.edu>
 220
 221 =head1 VERSION
 222
 223 $Id: validate_gff3.pl,v 1.1 2007/12/03 14:20:23 canaran Exp $
 224
 225 =head1 CREDITS
 226
 227 - SQLite support adapted from patch contributed by Robert Buels <rmb32@cornell.edu>.
 228
 229 =head1 COPYRIGHT AND LICENSE
 230
 231 Copyright (c) 2006-2007 Cold Spring Harbor Laboratory
 232
 233 This program is free software; you can redistribute it and/or modify it
 234 under the same terms as Perl itself. See DISCLAIMER.txt for
 235 disclaimers of warranty.
 236
 237 =cut
 238
 239 1;