Inheritance diagram for Bio::EnsEMBL::Utils::ConversionSupport:

Public Member Functions
public Bio::EnsEMBL::Utils::ConversionSupport	new ()

public Boolean	parse_common_options ()

public Boolean	parse_extra_options ()

public Array	allowed_params ()

public Array	get_common_params ()

public Array	get_loutre_params ()

public Array	get_annotrack_params ()

public void	remove_vega_params ()

public Boolean	confirm_params ()

public String	list_all_params ()

public String	create_commandline_options ()

public Boolean	check_required_params ()

public Boolean	user_proceed ()

public String	read_user_input ()

public Boolean	comma_to_list ()

public Boolean	list_or_file ()

public Scalar	param ()

public String	error ()

public Int	warnings ()

public String	serverroot ()

public DBAdaptor	get_database ()

public DBI	get_dbconnection ()

public Bio::EnsEMBL::DBSQL::DBAdaptor	dba ()

public Boolean	dynamic_use ()

public Hashref	get_chrlength ()

public Hashref	get_ensembl_chr_mapping ()

public Int	get_taxonomy_id ()

public String	get_species_scientific_name ()

public String	species ()

public List	sort_chromosomes ()

protected Values	_by_chr_num ()


public Hashref	split_chromosomes_by_size ()

public Boolean	is_patch ()

public Boolean	log ()

public	lock_log ()

public	unlock_log ()

public Boolean	log_warning ()

public void	log_error ()

public Boolean	log_verbose ()

public Boolean	log_stamped ()

public Filehandle	log_filehandle ()

public Filehandle	filehandle ()

public Filehandle	init_log_date ()

public Filehandle	init_log ()

public Boolean	finish_log ()

public String	date_and_mem ()

public String	date ()

public String	date_and_time ()

public	date_format ()

public String	mem ()

public A	commify ()

public Arrayref	fetch_non_hidden_slices ()

public Arrayref	get_non_hidden_slice_names ()

public Arrayref	get_wanted_chromosomes ()

public Boolean	is_haplotype ()

public Arrayref	get_unique_genes ()

public	get_attrib_values ()

public Arrayref	fix_attrib_value ()

protected Attrib_type	_get_attrib_id ()

public Array_ref	store_new_attribute ()

public Array_ref	update_attribute ()

public void	remove_duplicate_attribs ()

public	save_seq ()

public String	get_alignment ()

public	allowed_duplicate_regions ()

Detailed Description

Synopsis

my $serverroot = '/path/to/ensembl';
my $support = new Bio::EnsEMBL::Utils::ConversionSupport($serverroot);
 
# parse common options
$support->parse_common_options;
 
# parse extra options for your script
$support->parse_extra_options( 'string_opt=s', 'numeric_opt=n' );
 
# ask user if he wants to run script with these parameters
$support->confirm_params;
 
# see individual method documentation for more stuff

Description

This module is a collection of common methods and provides helper
functions for the Vega release and schema conversion scripts. Amongst
others, it reads options from a config file, parses commandline options
and does logging.

Definition at line 39 of file ConversionSupport.pm.

Member Function Documentation

◆ _by_chr_num()

protected Values Bio::EnsEMBL::Utils::ConversionSupport::_by_chr_num ( )

  Example     :

my @sorted = sort _by_chr_num qw(X, 6-COX, 14, 7);

  Description : Subroutine to use in sort for sorting chromosomes. Sorts
                numerically, then alphabetically
  Return type : values to be used by sort
  Exceptions  : none
  Caller      : internal ($self->sort_chromosomes)

Code:

click to view

◆ _get_attrib_id()

protected Attrib_type Bio::EnsEMBL::Utils::ConversionSupport::_get_attrib_id ( )

  Arg[1]      : attrib_type.code 
  Arg[2]      : database handle 
  Example     :

$self->_get_attrib_id('name',$dbh)

  Description : get attrib_type.attrib_type_id from a attrib_type.code
  Return type : attrib_type.attrib_type_id 
  Caller      : internal
  Status      : stable

Code:

click to view

◆ allowed_duplicate_regions()

public Bio::EnsEMBL::Utils::ConversionSupport::allowed_duplicate_regions ( )

Undocumented method

Code:

click to view

sub allowed_duplicate_regions {
  my $self = shift;
 
  # set up lists of vega names of wanted haplotype / strains
  my $regions = {
    'human' => [
      {
        '6'      => '28000000:34000000',
        '6-COX'  => 'all',
        '6-QBL'  => 'all',
        '6-APD'  => 'all',
        '6-MANN' => 'all',
        '6-MCF'  => 'all',
        '6-DBB'  => 'all',
        '6-SSTO' => 'all',
      },
      {
        '19'       => '54020000:54910000',
        '19-PGF_1' => 'all',
        '19-PGF_2' => 'all',
        '19-COX_1' => 'all',
        '19-COX_2' => 'all',
        '19-DM1A'  => 'all',
        '19-DM1B'  => 'all',
        '19-MC1A'  => 'all',
        '19-MC1B'  => 'all',
      },
    ],
    'mouse' => [
      {
        '1'               => '60564732:63711641',
        '1-Idd5.1_DIL'    => 'all',
        '1-Idd5.1_CHO'    => 'all',
      },
      {
        '1'               => '65533102:69307244',
        '1-Idd5.3_DIL'    => 'all',
      },
      {
        '1'               => '130232728:130661594',
        '1-Idd5.4_DIL'    => 'all',
      },
      {
        '3'               => '36492618:37600833',
        '3-Idd3_DIL'      => 'all',
        '3-Idd3_129'      => 'all',
      },
      {
        '3'               => '99848826:101467080',
        '3-Idd10_DIL'     => 'all',
      },
      {
        '3'               => '109144756:109930492',
        '3-Idd18.1_DIL'   => 'all',
      },
      {
        '3'               => '103489414:104054885',
        '3-Idd18.2_DIL'   => 'all',
      },
      {
        '4'               => '128371876:131853368',
        '4-Idd9.1_DIL'    => 'all',
      },
      {
        '4'               => '134841437:135252443',
        '4-Idd9.1M_DIL'   => 'all',
      },
      {
        '4'               => '146049976:149895141',
        '4-Idd9.2_DIL'    => 'all',
      },
      {
        '4'               => '149556939:151385803',
        '4-Idd9.3_DIL'    => 'all',
      },
      {
        '6'               => '143550839:149565172',
        '6-Idd6.1_2_CHO'  => 'all',
      },
      {
        '6'               => '129593784:131241919',
        '6-Idd6.AM_CHO'   => 'all',
      },
      {
        '11'              => '69890356:71340035',
        '11-Idd4.1_DIL'   => 'all'
      },
      {
        '11'              => '72734492:74404570',
        '11-Idd4.2_DIL'   => 'all',
      },    
      {
        '11'              => '86785996:90007691',
        '11-Idd4.2Q_CHO'  => 'all',
      },
      {
        '17'              => '27302611:29220265',
        '17-Idd16.1_CHO'  => 'all',
      },
      {
        '17'              => '33567721:38721962',
        '17-Idd1_CHO'     => 'all',
        '17-Idd1_DIL'     => 'all',
      },
    ],
    'pig' => [
      {
        '7'               => '24728583:29807435',
        '7-LW'            => 'all',
      },
      {
        'X'               => 'all',
        'X-WTSI'          => 'all',
      },
    ],
    'zebrafish' => [],
    'gorilla' => [],
    'wallaby' => [],
    'chimp' => [],
    'rat'   => [],
  };
 
return $regions;
 
}

◆ allowed_params()

public Array Bio::EnsEMBL::Utils::ConversionSupport::allowed_params ( )

  Arg[1-N]    : (optional) List of allowed parameters to set
  Example     :

my @allowed = $self->allowed_params(qw(param1 param2));

  Description : Getter/setter for allowed parameters. This is used by
                $self->confirm_params() to avoid cluttering of output with
                conffile entries not relevant for a given script. You can use
                $self->get_common_params() as a shortcut to set them.
  Return type : Array - list of allowed parameters
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ check_required_params()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::check_required_params ( )

  Arg[1-N]    : List @params - parameters to check
  Example     :

$self->check_required_params(qw(dbname host port));

  Description : Checks $self->param to make sure the requested parameters
                have been set. Dies if parameters are missing.
  Return type : true on success
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ comma_to_list()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::comma_to_list ( )

  Arg[1-N]    : list of parameter names to parse
  Example     :

$support->comma_to_list('chromosomes');

  Description : Transparently converts comma-separated lists into arrays (to
                allow different styles of commandline options, see perldoc
                Getopt::Long for details). Parameters are converted in place
                (accessible through $self->param('name')).
  Return type : true on success
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ commify()

public A Bio::EnsEMBL::Utils::ConversionSupport::commify ( )

  Arg[1]      : Int $num - a number to commify
  Example     :

print "An easy to read number: ".$self->commify(100000000);

@section autotoc_md31 will print 100,000,000

  Description : put commas into a number to make it easier to read
  Return type : a string representing the commified number
  Exceptions  : none
  Caller      : general
  Status      : stable

Code:

click to view

◆ confirm_params()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::confirm_params ( )

  Example     :

$support->confirm_params;

  Description : Prints a table of parameters that were collected from config
                file and commandline and asks user to confirm if he wants
                to proceed.
  Return type : true on success
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ create_commandline_options()

public String Bio::EnsEMBL::Utils::ConversionSupport::create_commandline_options ( )

  Arg[1]      : Hashref $settings - hashref describing what to do
                Allowed keys:
                    allowed_params => 0|1   # use all allowed parameters
                    exclude => []           # listref of parameters to exclude
                    replace => {param => newval} # replace value of param with
                                                 # newval
  Example     :

$support->create_commandline_options({
                   allowed_params => 1,
                   exclude => ['verbose'],
                   replace => { 'dbname' => 'homo_sapiens_vega_33_35e' }
               });

  Description : Creates a commandline options string that can be passed to any
                other script using ConversionSupport.
  Return type : String - commandline options string
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ date()

public String Bio::EnsEMBL::Utils::ConversionSupport::date ( )

  Example     :

print "Date: " . $support->date . "\n";

  Description : Prints a nicely formatted datetamp (YYYY-DD-MM)
  Return type : String - the timestamp
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ date_and_mem()

public String Bio::EnsEMBL::Utils::ConversionSupport::date_and_mem ( )

  Example     :

print LOG "Time, memory usage: ".$support->date_and_mem."\n";

  Description : Prints a timestamp and the memory usage of your script.
  Return type : String - timestamp and memory usage
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ date_and_time()

public String Bio::EnsEMBL::Utils::ConversionSupport::date_and_time ( )

  Example     :

print "Date: " . $support->date . "\n";

  Description : Prints a nicely formatted timestamp (YYYY-DD-MM hh:mm:ss)
  Return type : String - the timestamp
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ date_format()

public Bio::EnsEMBL::Utils::ConversionSupport::date_format ( )

Undocumented method

Code:

click to view

◆ dba()

public Bio::EnsEMBL::DBSQL::DBAdaptor Bio::EnsEMBL::Utils::ConversionSupport::dba ( )

  Arg[1]      : (optional) String $database - type of db apaptor to retrieve
  Example     :

my $dba = $support->dba;

  Description : Getter for database adaptor. Returns default (i.e. created
                first) db adaptor if no argument is provided.
  Return type : Bio::EnsEMBL::DBSQL::DBAdaptor or Bio::Otter::DBSQL::DBAdaptor
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ dynamic_use()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::dynamic_use ( )

  Arg [1]    : String $classname - The name of the class to require/import
  Example    :

$self->dynamic_use('Bio::EnsEMBL::DBSQL::DBAdaptor');

  Description: Requires and imports the methods for the classname provided,
               checks the symbol table so that it doesnot re-require modules
               that have already been required.
  Returntype : true on success
  Exceptions : Warns to standard error if module fails to compile
  Caller     : internal

Code:

click to view

◆ error()

public String Bio::EnsEMBL::Utils::ConversionSupport::error ( )

  Arg[1]      : (optional) String - error message
  Example     :

$support->error("An error occurred: $@");

exit(0) if $support->error;

  Description : Getter/setter for error messages
  Return type : String - error message
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ fetch_non_hidden_slices()

public Arrayref Bio::EnsEMBL::Utils::ConversionSupport::fetch_non_hidden_slices ( )

  Arg[1]      : B::E::SliceAdaptor
  Arg[2]      : B::E::AttributeAdaptor
  Arg[3]      : string $coord_system_name (optional) - 'chromosome' by default
  Arg[4]      : string $coord_system_version (optional) - 'otter' by default
  Example     :

$chroms = $support->fetch_non_hidden_slice($sa,$aa);

  Description : retrieve all slices from a loutre database that don't have a hidden attribute.
                Doesn't retrieve non-reference slices
  Return type : arrayref
  Caller      : general
  Status      : stable

Code:

click to view

◆ filehandle()

public Filehandle Bio::EnsEMBL::Utils::ConversionSupport::filehandle ( )

  Arg[1]      : String $mode - file access mode
  Arg[2]      : String $file - input or output file
  Example     :

 my $fh = $support->filehandle('>>', '/path/to/file');
@section autotoc_md30 print to the filehandle
                print $fh 'Your text goes here...\n';

  Description : Returns a filehandle (*STDOUT for writing, *STDIN for reading
                by default) to print to or read from.
  Return type : Filehandle - the filehandle
  Exceptions  : thrown if file can't be opened
  Caller      : general

Code:

click to view

◆ finish_log()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::finish_log ( )

  Example     :

$support->finish_log;

  Description : Writes footer information to a logfile. This includes the
                number of logged warnings, timestamp and memory footprint.
  Return type : TRUE on success
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ fix_attrib_value()

public Arrayref Bio::EnsEMBL::Utils::ConversionSupport::fix_attrib_value ( )

  Arg[1]      : Arrayref of existing B::E::Attributes
  Arg[2]      : dbID of object
  Arg[3]      : name of object (just for reporting)
  Arg[4]      : attrib_type.code
  Arg[5]      : attrib_type.value
  Arg[6]      : interactive ? (0 by default)
  Arg[7]      : table
  Example     :

$support->fix_attrib_value($attribs,$chr_id,$chr_name,'vega_export_mod','N',1);

  Description : adds a new attribute to an object, or updates an existing attribute with a new value
                Can be run in interactive or non-interactive mode (default)
  Return type : arrayref of results
  Caller      : general
  Status      : only ever tested with seq_region_attributes to date

Code:

click to view

◆ get_alignment()

public String Bio::EnsEMBL::Utils::ConversionSupport::get_alignment ( )

  Arg[1]      : string (first sequence)
  Arg[1]      : string (second sequence)
  Arg[1]      : string (sequence type))
  Example     :

$support->get_alignment('AAAAA','CCCCCCC','DNA')

  Description : creates a temporary file containing the sequence you give it
  Return type : string (filename)
  Caller      : general
  Status      : stable

Code:

click to view

sub get_alignment {
  my $self = shift;
  my $ext_seq  = shift || return undef;
  my $int_seq  = shift || return undef;
  $int_seq =~ s/<br \/>//g;
  my $seq_type = shift || return undef;
 
 
  # To stop box running out of memory - put an upper limit on the size of sequence
  # that alignview can handle
  if (length $int_seq > 1e6 || length $ext_seq > 1e6)  {
    $self->log_error('Cannot align if sequence > 1 Mbase');
  }
 
  my $int_seq_file = $self->save_seq($int_seq);
  my $ext_seq_file = $self->save_seq($ext_seq);
 
  my $label_width  = '22'; # width of column for e! object label
  my $output_width = 61;   # width of alignment
  my $dnaAlignExe  = '/localsw/bin/emboss/bin/matcher -asequence %s -bsequence %s -outfile %s';
  my $pepAlignExe  = '/localsw/bin/wise2/bin/psw -dymem explicit -m /localsw/bin/wise2/wisecfg/blosum62.bla %s %s -n %s -w %s > %s';
 
  my $out_file = time() . int(rand()*100000000) . $$;
  $out_file = $self->param('logpath').'/' . $out_file . '.out';
 
  my $command;
  my $fh;
  if ($seq_type eq 'DNA') {
    $command = sprintf $dnaAlignExe, $int_seq_file, $ext_seq_file, $out_file;
    `$command`;
    unless (open($fh, "<", $out_file)) {
      $command = sprintf $dnaAlignExe, $int_seq_file, $ext_seq_file, $out_file;
      `$command`;
    }
  }
  elsif ($seq_type eq 'PEP') {
    $command = sprintf $pepAlignExe, $int_seq_file, $ext_seq_file, $label_width, $output_width, $out_file;
    `$command`;
    unless (open($fh, "<", $out_file)) {
      $self->log_warning("Cannot open alignment file\n");
    }
  }
  my $alignment ;
  while (<$fh>) {
    next if $_ =~ 
    /\#Report_file
     |\#----.*
     |\/\/\s*
     |\#\#\#
     |^\#$
     |Rundate: #matcher
     |Commandline #matcher
     |asequence #matcher
     |bsequence #matcher
     |outfile #matcher
     |aformat #matcher
     |Align_format #matcher
     |Report_file #matcher
     /x;
    $alignment .= $_;
  }
 
  $alignment =~ s/\n+$//;
  unlink $out_file;
  unlink $int_seq_file;
  unlink $ext_seq_file;
  return $alignment;
}

◆ get_annotrack_params()

public Array Bio::EnsEMBL::Utils::ConversionSupport::get_annotrack_params ( )

  Arg         : (optional) return a list to parse or not
  Example     :

$support->parse_extra_options($support->get_annotrack_params('parse'))

  Description : Returns a list of parameters used to connect to annotrack
  Return type : Array - list of common parameters
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ get_attrib_values()

public Bio::EnsEMBL::Utils::ConversionSupport::get_attrib_values ( )

Undocumented method

Code:

click to view

◆ get_chrlength()

public Hashref Bio::EnsEMBL::Utils::ConversionSupport::get_chrlength ( )

  Arg[1]      : (optional) Bio::EnsEMBL::DBSQL::DBAdaptor $dba
  Arg[2]      : (optional) String $version - coord_system version
  Arg[3]      : (optional) String $type - type of region eg chromsome (defaults to 'toplevel')
  Arg[4]      : (optional) Boolean - return non reference slies as well (required for haplotypes eq 6-COX)
  Arg[5]      : (optional) Override chromosome parameter filtering with this array reference. Empty denotes all.
  Example     :

my $chr_length = $support->get_chrlength($dba);

  Description : Get all chromosomes and their length from the database. Return
                chr_name/length for the chromosomes the user requested (or all
                chromosomes by default)
  Return type : Hashref - chromosome_name => length
  Exceptions  : thrown if not passing a Bio::EnsEMBL::DBSQL::DBAdaptor
  Caller      : general

Code:

click to view

◆ get_common_params()

public Array Bio::EnsEMBL::Utils::ConversionSupport::get_common_params ( )

  Example     :

my @allowed_params = $self->get_common_params, 'extra_param';

  Description : Returns a list of commonly used parameters in the conversion
                scripts. Shortcut for setting allowed parameters with
                $self->allowed_params().
  Return type : Array - list of common parameters
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ get_database()

public DBAdaptor Bio::EnsEMBL::Utils::ConversionSupport::get_database ( )

  Arg[1]      : String $database - the type of database to connect to
                (eg core, otter)
  Arg[2]      : (optional) String $prefix - the prefix used for retrieving the
                connection settings from the configuration
  Example     :

my $db = $support->get_database('core');

  Description : Connects to the database specified.
  Return type : DBAdaptor of the appropriate type
  Exceptions  : thrown if asking for unknown database
  Caller      : general

Code:

click to view

◆ get_dbconnection()

public DBI Bio::EnsEMBL::Utils::ConversionSupport::get_dbconnection ( )

  Arg[1]      : (optional) String $prefix - the prefix used for retrieving the
                connection settings from the configuration
  Example     :

my $dbh = $self->get_dbconnection;

  Description : Connects to the database server specified. You don't have to
                specify a database name (this is useful for running commands
                like $dbh->do('show databases')).
  Return type : DBI database handle
  Exceptions  : thrown if connection fails
  Caller      : general
  Status      : At Risk

Code:

click to view

◆ get_ensembl_chr_mapping()

public Hashref Bio::EnsEMBL::Utils::ConversionSupport::get_ensembl_chr_mapping ( )

  Arg[1]      : (optional) Bio::EnsEMBL::DBSQL::DBAdaptor $dba
  Arg[2]      : (optional) String $version - coord_system version
  Example     :

my $ensembl_mapping = $support->get_ensembl_chr_mapping($dba);

  Description : Gets a mapping between Vega chromosome names and their
                equivalent Ensembl chromosomes. Works with non-reference chromosomes
  Return type : Hashref - Vega name => Ensembl name
  Exceptions  : thrown if not passing a Bio::EnsEMBL::DBSQL::DBAdaptor
  Caller      : general

Code:

click to view

◆ get_loutre_params()

public Array Bio::EnsEMBL::Utils::ConversionSupport::get_loutre_params ( )

  Arg         : (optional) return a list to parse or not
  Example     :

$support->parse_extra_options($support->get_loutre_params('parse'))

  Description : Returns a list of commonly used loutre db parameters - parse option is
                simply used to distinguish between reporting and parsing parameters
  Return type : Array - list of common parameters
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ get_non_hidden_slice_names()

public Arrayref Bio::EnsEMBL::Utils::ConversionSupport::get_non_hidden_slice_names ( )

  Arg[1]      : B::E::SliceAdaptor
  Arg[2]      : B::E::AttributeAdaptor
  Arg[3]      : string $coord_system_name (optional) - 'chromosome' by default
  Arg[4]      : string $coord_system_version (optional) - 'otter' by default
  Example     :

$chrom_names = $support->get_non_hidden_slice_names($sa,$aa);

  Description : retrieve names of all slices from a loutre database that don't have a hidden attribute.
                Doesn't retrieve non-reference slices
  Return type : arrayref of names of all non-hidden slices
  Caller      : general
  Status      : stable

Code:

click to view

◆ get_species_scientific_name()

public String Bio::EnsEMBL::Utils::ConversionSupport::get_species_scientific_name ( )

  Arg[1]      : Bio::EnsEMBL::DBSQL::DBAdaptor $dba
  Example     :

my $species = $support->get_species_scientific_name($dba);

  Description : Retrieves the species scientific name (Genus species) from the
                meta table
  Return type : String - species scientific name
  Exceptions  : thrown if species name can not be determined from db
  Caller      : general

Code:

click to view

◆ get_taxonomy_id()

public Int Bio::EnsEMBL::Utils::ConversionSupport::get_taxonomy_id ( )

  Arg[1]      : Bio::EnsEMBL::DBSQL::DBAdaptor $dba
  Example     :

my $sid = $support->get_taxonony_id($dba);

  Description : Retrieves the taxononmy ID from the meta table
  Return type : Int - the taxonomy ID
  Exceptions  : thrown if no taxonomy ID is found in the database
  Caller      : general

Code:

click to view

◆ get_unique_genes()

public Arrayref Bio::EnsEMBL::Utils::ConversionSupport::get_unique_genes ( )

  Arg[1]      : B::E::Slice
  Arg[2]      : B::E::DBAdaptor (optional, if you don't supply one then the *first* one you generated is returned, which may or may not be what you want!)
  Example     :

$genes = $support->get_unique_genes($slice,$dba);

  Description : Retrieve genes that are only on the slice itself - used for human where assembly patches
                are in the assembly_exception table. Needs the PATCHes to have 'non_ref' seq_region_attributes.
  Return type : arrayref of genes
  Caller      : general
  Status      : stable

Code:

click to view

◆ get_wanted_chromosomes()

public Arrayref Bio::EnsEMBL::Utils::ConversionSupport::get_wanted_chromosomes ( )

  Arg[1]      : B::E::SliceAdaptor
  Arg[2]      : B::E::AttributeAdaptor
  Arg[3]      : string $coord_system_name (optional) - 'chromosome' by default
  Arg[4]      : string $coord_system_version (optional) - 'otter' by default
  Example     :

$chr_names = $support->get_wanted_chromosomes($laa,$lsa);

  Description : retrieve names of slices from a lutra database that are ready for dumping to Vega.
                Deals with list of names to ignore (ignore_chr = LIST)
  Return type : arrayref of slices
  Caller      : general
  Status      : stable

Code:

click to view

◆ init_log()

public Filehandle Bio::EnsEMBL::Utils::ConversionSupport::init_log ( )

  Example     :

$support->init_log;

  Description : Opens a filehandle to the logfile and prints some header
                information to this file. This includes script name, date, user
                running the script and parameters the script will be running
                with.
  Return type : Filehandle - the log filehandle
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ init_log_date()

public Filehandle Bio::EnsEMBL::Utils::ConversionSupport::init_log_date ( )

  Example     :

$support->init_log_date;

  Description : Opens a filehandle to a logfile with the date in the file name
  Return type : Filehandle - the log filehandle
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ is_haplotype()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::is_haplotype ( )

  Arg[1]      : B::E::Slice
  Arg[2]:     : B::E::DBAdaptor (optional, if you don't supply one then the *first* one you generated is returned, which may or may not be what you want!)
  Description : Is the slice a Vega haplotype? At the moment this is 
    implemented by testing for presence of vega_ref_chrom but non_ref
    which is correct in practice, but really misses the prupose of
    vega_ref_chrom, so this might bite us if that changes.
  Return type : boolean

Code:

click to view

◆ is_patch()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::is_patch ( )

  Arg[1]      : B::E::Slice
  Example     :

if ($support->is_patch($slice)) { ...

  Description : Looks at seq_region attributes to decide if a slice is a patch or not
                If PATCH seq_region_attrib is not there check to see if name suggests it is a PATCH
  Return type : true/false

Code:

click to view

◆ list_all_params()

public String Bio::EnsEMBL::Utils::ConversionSupport::list_all_params ( )

  Example     :

print LOG $support->list_all_params;

  Description : prints a table of the parameters used in the script
  Return type : String - the table to print
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ list_or_file()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::list_or_file ( )

  Arg[1]      : Name of parameter to parse
  Example     :

$support->list_or_file('gene');

  Description : Determines whether a parameter holds a list or it is a filename
                to read the list entries from.
  Return type : true on success
  Exceptions  : thrown if list file can't be opened
  Caller      : general

Code:

click to view

◆ lock_log()

public Bio::EnsEMBL::Utils::ConversionSupport::lock_log ( )

  Description : Use flock-style locks to lock log and fastforward to end.
                Useful if log is being written to by multiple processes.

Code:

click to view

◆ log()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::log ( )

  Arg[1]      : String $txt - the text to log
  Arg[2]      : Int $indent - indentation level for log message
  Example     :

my $log = $support->log_filehandle;

$support->log('Log foo.\n', 1);

  Description : Logs a message to the filehandle initialised by calling
                $self->log_filehandle(). You can supply an indentation level
                to get nice hierarchical log messages.
  Return type : true on success
  Exceptions  : thrown when no filehandle can be obtained
  Caller      : general

Code:

click to view

◆ log_error()

public void Bio::EnsEMBL::Utils::ConversionSupport::log_error ( )

  Arg[1]      : String $txt - the error text to log
  Arg[2]      : Int $indent - indentation level for log message
  Example     :

my $log = $support->log_filehandle;

$support->log_error('Log foo.\n', 1);

  Description : Logs a message via $self->log and exits the script.
  Return type : none
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ log_filehandle()

public Filehandle Bio::EnsEMBL::Utils::ConversionSupport::log_filehandle ( )

  Arg[1]      : (optional) String $mode - file access mode
  Example     :

 my $log = $support->log_filehandle;
@section autotoc_md28 print to the filehandle
                print $log 'Lets start logging...\n';
@section autotoc_md29 log via the wrapper $self->log()
                $support->log('Another log message.\n');

  Description : Returns a filehandle for logging (STDERR by default, logfile if
                set from config or commandline). You can use the filehandle
                directly to print to, or use the smart wrapper $self->log().
                Logging mode (truncate or append) can be set by passing the
                mode as an argument to log_filehandle(), or with the
                --logappend commandline option (default: truncate)
  Return type : Filehandle - the filehandle to log to
  Exceptions  : thrown if logfile can't be opened
  Caller      : general

Code:

click to view

◆ log_stamped()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::log_stamped ( )

  Arg[1]      : String $txt - the warning text to log
  Arg[2]      : Int $indent - indentation level for log message
  Example     :

my $log = $support->log_filehandle;

$support->log_stamped('Log this stamped message.\n', 1);

  Description : Appends timestamp and memory usage to a message and logs it via
                $self->log
  Return type : TRUE on success
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ log_verbose()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::log_verbose ( )

  Arg[1]      : String $txt - the warning text to log
  Arg[2]      : Int $indent - indentation level for log message
  Example     :

my $log = $support->log_filehandle;

$support->log_verbose('Log this verbose message.\n', 1);

  Description : Logs a message via $self->log if --verbose option was used
  Return type : TRUE on success, FALSE if not verbose
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ log_warning()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::log_warning ( )

  Arg[1]      : String $txt - the warning text to log
  Arg[2]      : Int $indent - indentation level for log message
  Arg[3]      : Bool - add a line break before warning if true
  Example     :

my $log = $support->log_filehandle;

$support->log_warning('Log foo.\n', 1);

  Description : Logs a message via $self->log and increases the warning counter.
  Return type : true on success
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ mem()

public String Bio::EnsEMBL::Utils::ConversionSupport::mem ( )

  Example     :

print "Memory usage: " . $support->mem . "\n";

  Description : Prints the memory used by your script. Not sure about platform
                dependence of this call ...
  Return type : String - memory usage
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ new()

public Bio::EnsEMBL::Utils::ConversionSupport Bio::EnsEMBL::Utils::ConversionSupport::new ( )

  Arg[1]      : String $serverroot - root directory of your ensembl sandbox
  Example     :

my $support = new Bio::EnsEMBL::Utils::ConversionSupport(

'/path/to/ensembl');

  Description : constructor
  Return type : Bio::EnsEMBL::Utils::ConversionSupport object
  Exceptions  : thrown if no serverroot is provided
  Caller      : general

Code:

click to view

◆ param()

public Scalar Bio::EnsEMBL::Utils::ConversionSupport::param ( )

  Arg[1]      : Parameter name
  Arg[2-N]    : (optional) List of values to set
  Example     :

my $dbname = $support->param('dbname');
               $support->param('port', 3306);
               $support->param('chromosomes', 1, 6, 'X');

  Description : Getter/setter for parameters. Accepts single-value params and
                list params.
  Return type : Scalar value for single-value parameters, array of values for
                list parameters
  Exceptions  : thrown if no parameter name is supplied
  Caller      : general

Code:

click to view

◆ parse_common_options()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::parse_common_options ( )

  Example     :

$support->parse_common_options;

  Description : This method reads options from a configuration file and parses
                some commandline options that are common to all scripts (like
                db connection settings, help, dry-run). Commandline options
                will override config file settings.

                All options will be accessible via $self->param('name').
  Return type : true on success 
  Exceptions  : thrown if configuration file can't be opened
  Caller      : general

Code:

click to view

sub parse_common_options {
  my $self = shift;
 
  # read commandline options
  my %h;
  Getopt::Long::Configure("pass_through");
  &GetOptions( \%h,
           'dbname|db_name=s',
           'host|dbhost|db_host=s',
           'port|dbport|db_port=n',
           'user|dbuser|db_user=s',
           'pass|dbpass|db_pass=s',
               'prod_dbname=s',
               'prod_host=s',
               'prod_port=n',
               'prod_user=s',
               'prod_pass=s',
           'conffile|conf=s',
           'logfile|log=s',
               'nolog',
           'logpath=s',
               'log_base_path=s',
           'logappend|log_append',
           'verbose|v',
           'interactive|i=s',
         'hideparamlist=s',
         'joblog=s',
           'dry_run|dry|n',
           'help|h|?',
         );
 
  # reads config file
  my $conffile = $h{'conffile'} || $self->serverroot . "/sanger-plugins/vega/conf/ini-files/Conversion.ini";
  $conffile = abs_path($conffile);
  if (-e $conffile) {
    open(my $fh, '<', $conffile) or throw( 
      "Unable to open configuration file $conffile for reading: $!");
    my $serverroot = $self->serverroot;
    while (<$fh>) {
      chomp;
 
      # remove comments
      s/^[#;].*//;
      s/\s+[;].*$//;
 
      # read options into internal parameter datastructure, removing whitespace
      next unless (/(\w\S*)\s*=\s*(\S*)\s*/);
      my $name = $1;
      my $val = $2;
      if ($val =~ /\$SERVERROOT/) {
    $val =~ s/\$SERVERROOT/$serverroot/g;
    $val = abs_path($val);
      }
      $self->param($name, $val);
    }
    close $fh;
    $self->param('conffile', $conffile);
  }
  elsif ($conffile) {
    warning("Unable to open configuration file $conffile for reading: $!");
  }
 
# override configured parameter with commandline options
  map { $self->param($_, $h{$_}) } keys %h;
 
  return (1) if $self->param('nolog');
 
  # if logpath & logfile are not set, set them here to /ensemblweb/vega_dev/shared/logs/conversion/DBNAME/SCRIPNAME_NN.log
  if (! defined($self->param('log_base_path')))  {
    $self->param('log_base_path','/ensemblweb/shared/logs/conversion/');
  }
  my $dbname = $self->param('dbname');
  $dbname =~ s/^vega_//;
  if (not (defined($self->param('logpath')) )){
    $self->param('logpath', $self->param('log_base_path')."/".$dbname."/" );
  }
  if ( not defined $self->param('logfile') ){
    my $log = $Script;
    $log =~ s/.pl$//g;
    my $counter;
    for ($counter=1 ; (-e $self->param('logpath')."/".$log."_".sprintf("%03d", $counter).".log"); $counter++){
      #      warn  $self->param('logpath')."/".$log."_".$counter.".log";
    }
    $self->param('logfile', $log."_".sprintf("%03d", $counter).".log");
  }
  return(1);
}

◆ parse_extra_options()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::parse_extra_options ( )

  Arg[1-N]    : option descriptors that will be passed on to Getopt::Long
  Example     :

$support->parse_extra_options('string_opt=s', 'numeric_opt=n');

  Description : Parse extra commandline options by passing them on to
                Getopt::Long and storing parameters in $self->param('name).
                If the last option is '...' then unknown options are kept
                in @ARGV for future calls to this method. Without '...'
                then it is an error to have further options.
  Return type : true on success
  Exceptions  : none (caugth by $self->error)
  Caller      : general

Code:

click to view

◆ read_user_input()

public String Bio::EnsEMBL::Utils::ConversionSupport::read_user_input ( )

  Arg[1]      : (optional) String $text - notification text to present to user
  Example     :

 my $ret = $support->read_user_input("Choose a number [1/2/3]");
                if ($ret == 1) {
@section autotoc_md26 do something
                } elsif ($ret == 2) {
@section autotoc_md27 do something else
                }

  Description : If running interactively, the user is asked for input.
  Return type : String - user's input
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ remove_duplicate_attribs()

public void Bio::EnsEMBL::Utils::ConversionSupport::remove_duplicate_attribs ( )

  Arg[1]      : db handle
  Arg[2]      : table
  Example     :

$support->remove_duplicate_attribs($dbh,'gene');

  Description : uses MySQL to remove duplicate entries from an attribute table
  Return type : none
  Caller      : general
  Status      : stable

Code:

click to view

◆ remove_vega_params()

public void Bio::EnsEMBL::Utils::ConversionSupport::remove_vega_params ( )

  Example     :

$support->remove_vega_params;

  Description : Removes Vega db conection parameters. Usefull to avoid clutter in log files when
                working exclusively with loutre
  Return type : none
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ save_seq()

public Bio::EnsEMBL::Utils::ConversionSupport::save_seq ( )

Undocumented method

Code:

click to view

◆ serverroot()

public String Bio::EnsEMBL::Utils::ConversionSupport::serverroot ( )

  Arg[1]      : (optional) String - root directory of your ensembl sandbox
  Example     :

my $serverroot = $support->serverroot;

  Description : Getter/setter for the root directory of your ensembl sandbox.
                This is set when ConversionSupport object is created, so
                usually only used as a getter.
  Return type : String - the server root directory
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ sort_chromosomes()

public List Bio::EnsEMBL::Utils::ConversionSupport::sort_chromosomes ( )

  Arg[1]      : (optional) Hashref $chr_hashref - Hashref with chr_name as keys
  Example     :

my $chr = { '6-COX' => 1, '1' => 1, 'X' => 1 };

my @sorted = $support->sort_chromosomes($chr);

  Description : Sorts chromosomes in an intuitive way (numerically, then
                alphabetically). If no chromosome hashref is passed, it's
                retrieve by calling $self->get_chrlength()
  Return type : List - sorted chromosome names
  Exceptions  : thrown if no hashref is provided
  Caller      : general

Code:

click to view

◆ species()

public String Bio::EnsEMBL::Utils::ConversionSupport::species ( )

  Arg[1]      : (optional) String $species - species name to set
  Example     :

my $species = $support->species;

                my $url = "http://vega.sanger.ac.uk/$species/";
  Description : Getter/setter for species name (Genus_species). If not set, it's
                determined from database's meta table
  Return type : String - species name
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ split_chromosomes_by_size()

public Hashref Bio::EnsEMBL::Utils::ConversionSupport::split_chromosomes_by_size ( )

  Arg[1]      : (optional) Int $cutoff - the cutoff in bp between small and
                large chromosomes
  Arg[2]      : (optional) Boolean to include duplicate regions, ie PAR or not
                (default is no)
  Arg[3]      : (optional) Coordsystem version to retrieve

  Example     :

my $chr_slices = $support->split_chromosomes_by_size;

foreach my $block_size (keys %{ $chr_slices }) {

                    print "Chromosomes with blocksize $block_size: ";
                    print join(", ", map { $_->seq_region_name }
 $chr_slices->{$block_size} });
                }
  Description : Determines block sizes for storing DensityFeatures on
                chromosomes, and return slices for each chromosome. The block
                size is determined so that you have 150 bins for the smallest
                chromosome over 5 Mb in length. For chromosomes smaller than 5
                Mb, an additional smaller block size is used to yield 150 bins
                for the overall smallest chromosome. This will result in
                reasonable resolution for small chromosomes and high
                performance for big ones. Does not return non-reference seq_regions
  Return type : Hashref (key: block size; value: Arrayref of chromosome
                Bio::EnsEMBL::Slices)
  Exceptions  : none
  Caller      : density scripts

Code:

click to view

sub split_chromosomes_by_size {
  my $self   = shift;
  my $cutoff = shift || 5000000;
  my $dup    = shift || 0;
  my $cs_version = shift;
  my $include_non_reference = 1; #get non reference slices
  my $slice_adaptor = $self->dba->get_SliceAdaptor;
  my ($top_slices, $wanted_slices, $missed_slices);
  if ($self->param('chromosomes')) {
    foreach my $chr ($self->param('chromosomes')) {
      push @{ $top_slices }, $slice_adaptor->fetch_by_region('chromosome', $chr);
    }
  }
  else {
    $top_slices = $slice_adaptor->fetch_all('chromosome',$cs_version,$include_non_reference,$dup);
  }
 
  # filter out patches, if present
  $wanted_slices = [ grep { $_->is_reference or $self->is_haplotype($_,$self->dba) } @$top_slices ];
 
  # make a note of which ones are excluded
  $missed_slices = [ grep { ! $_->is_reference and ! $self->is_haplotype($_,$self->dba) } @$top_slices ];
 
  #warn non PATCH toplevels slice that are excluded (could be haplotypes if an earlier stage has failed)
  foreach (@{$missed_slices || []}) {
    my $sr_name = $_->seq_region_name;
    if ($self->is_patch($_)) {
      $self->log("Excluding $sr_name\n");
    }
    else {
      $self->log_warning("Excluding $sr_name, are you sure ?\n");
    }
  }
 
  my ($big_chr, $small_chr, $min_big_chr, $min_small_chr);
  foreach my $slice (@{ $wanted_slices }) {
    next if ($slice->length eq 10000); #hack for chrY pseudoslice
    if ($slice->length < $cutoff) {
      if (! $min_small_chr or ($min_small_chr > $slice->length)) {
    $min_small_chr = $slice->length;
      }
      # push small chromosomes onto $small_chr
      push @{ $small_chr }, $slice;
    }
    elsif (! $min_big_chr or ($min_big_chr > $slice->length) ){
      $min_big_chr = $slice->length;
    }
    # push _all_ chromosomes onto $big_chr
    push @{ $big_chr }, $slice;
  }
  my $chr_slices;
  $chr_slices->{int($min_big_chr/150)}   = $big_chr if $min_big_chr;
  $chr_slices->{int($min_small_chr/150)} = $small_chr if $min_small_chr;
  return $chr_slices;
}

◆ store_new_attribute()

public Array_ref Bio::EnsEMBL::Utils::ConversionSupport::store_new_attribute ( )

  Arg[1]      : seq_region.seq_region_id
  Arg[2]      : attrib_type.code
  Arg[3]      : attrib_type.value
  ARG[4]      : table to update (seq_region_attribute by default)
  Example     :

$support->store_new_attribute(23,name,5);

  Description : uses MySQL to store an entry (code and value) in an attribute table 
                (seq_region_attrib by default)
  Return type : array_ref
  Caller      : general
  Status      : stable

Code:

click to view

◆ unlock_log()

public Bio::EnsEMBL::Utils::ConversionSupport::unlock_log ( )

  Description : Unlock log previously locked by lock_log.

Code:

click to view

◆ update_attribute()

public Array_ref Bio::EnsEMBL::Utils::ConversionSupport::update_attribute ( )

  Arg[1]      : seq_region.seq_region_id
  Arg[2]      : attrib_type.code
  Arg[3]      : attrib_type.value
  ARG[4]      : table to update (seq_region_attribute by default)
  Example     :

$support->update_attribute(23,name,5);

  Description : uses MySQL to update an attribute table (seq_region_attrib by default)
  Return type : array_ref
  Caller      : general
  Status      : stable

Code:

click to view

◆ user_proceed()

public Boolean Bio::EnsEMBL::Utils::ConversionSupport::user_proceed ( )

  Arg[1]      : (optional) String $text - notification text to present to user
  Example     :

 # run a code snipped conditionally
                if ($support->user_proceed("Run the next code snipped?")) {
@section autotoc_md24 run some code
                }
 
@section autotoc_md25 exit if requested by user
                exit unless ($support->user_proceed("Want to continue?"));

  Description : If running interactively, the user is asked if he wants to
                perform a script action. If he doesn't, this section is skipped
                and the script proceeds with the code. When running
                non-interactively, the section is run by default.
  Return type : TRUE to proceed, FALSE to skip.
  Exceptions  : none
  Caller      : general

Code:

click to view

◆ warnings()

public Int Bio::EnsEMBL::Utils::ConversionSupport::warnings ( )

  Example     :

print LOG "There were ".$support->warnings." warnings.\n";

  Description : Returns the number of warnings encountered while running the
                script (the warning counter is increased by $self->log_warning).
  Return type : Int - number of warnings
  Exceptions  : none
  Caller      : general

Code:

click to view

The documentation for this class was generated from the following file:

ensembl/modules/Bio/EnsEMBL/Utils/ConversionSupport.pm

Public Member Functions

Detailed Description

Synopsis

Description

Member Function Documentation

◆ _by_chr_num()

◆ _get_attrib_id()

◆ allowed_duplicate_regions()

◆ allowed_params()

◆ check_required_params()

◆ comma_to_list()

◆ commify()

◆ confirm_params()

◆ create_commandline_options()

◆ date()

◆ date_and_mem()

◆ date_and_time()

◆ date_format()

◆ dba()

◆ dynamic_use()

◆ error()

◆ fetch_non_hidden_slices()

◆ filehandle()

◆ finish_log()

◆ fix_attrib_value()

◆ get_alignment()

◆ get_annotrack_params()

◆ get_attrib_values()

◆ get_chrlength()

◆ get_common_params()

◆ get_database()

◆ get_dbconnection()

◆ get_ensembl_chr_mapping()

◆ get_loutre_params()

◆ get_non_hidden_slice_names()

◆ get_species_scientific_name()

◆ get_taxonomy_id()

◆ get_unique_genes()

◆ get_wanted_chromosomes()

◆ init_log()

◆ init_log_date()

◆ is_haplotype()

◆ is_patch()

◆ list_all_params()

◆ list_or_file()

◆ lock_log()

◆ log()

◆ log_error()

◆ log_filehandle()

◆ log_stamped()

◆ log_verbose()

◆ log_warning()

◆ mem()

◆ new()

◆ param()

◆ parse_common_options()

◆ parse_extra_options()

◆ read_user_input()

◆ remove_duplicate_attribs()

◆ remove_vega_params()

◆ save_seq()

◆ serverroot()

◆ sort_chromosomes()

◆ species()

◆ split_chromosomes_by_size()

◆ store_new_attribute()

◆ unlock_log()

◆ update_attribute()

◆ user_proceed()

◆ warnings()