view release on metacpan or  search on metacpan

lib/CLIPSeqTools/CompareApp/libraries_overlap_stats.pm  view on Meta::CPAN

  Input options for library.
    --driver <Str>          driver for database connection (eg. mysql,
                            SQLite).
    --database <Str>        database name or path to database file for
                            file based databases (eg. SQLite).
    --table <Str>           database table.
    --host <Str>            hostname for database connection.
    --user <Str>            username for database connection.
    --password <Str>        password for database connection.
    --records_class <Str>   type of records stored in database.
    --filter <Filter>       filter library. May be used multiple times.
                            Syntax: column_name="pattern"
                            e.g. keep reads with deletions AND not repeat
                                 masked AND longer than 31
                                 --filter deletion="def" 
                                 --filter rmsk="undef" .
                                 --filter query_length=">31".
                            Operators: >, >=, <, <=, =, !=, def, undef

  Input options for reference library.
    --r_driver <Str>        driver for database connection (eg. mysql,
                            SQLite).
    --r_database <Str>      database name or path to database file for
                            file based databases (eg. SQLite).
    --r_table <Str>         database table.
    --r_host <Str>          hostname for database connection.
    --r_user <Str>          username for database connection.
    --r_password <Str>      password for database connection.
    --r_records_class <Str> type of records stored in database.
    --r_filter <Filter>     same as filter but for reference library.

  Other input.
    --rname_sizes <Str>    file with sizes for reference alignment
                           sequences (rnames). Must be tab delimited
                           (chromosome\tsize) with one line per rname.

  Output
    --o_prefix <Str>       output path prefix. Script will create and add
                           extension to path. [Default: ./]

  Other options.
    -v --verbose           print progress lines and extra information.
    -h -? --usage --help   print help message

=cut

package CLIPSeqTools::CompareApp::libraries_overlap_stats;
$CLIPSeqTools::CompareApp::libraries_overlap_stats::VERSION = '1.0.0';

# Make it an app command
use MooseX::App::Command;
extends 'CLIPSeqTools::CompareApp';


#######################################################################
#######################   Load External modules   #####################
#######################################################################
use Modern::Perl;
use autodie;
use namespace::autoclean;
use PDL::Lite; $PDL::BIGPDL = 0; $PDL::BIGPDL++; # enable huge pdls


#######################################################################
#######################   Command line options   ######################
#######################################################################
option 'rname_sizes' => (
	is            => 'rw',
	isa           => 'Str',
	required      => 1,
	documentation => 'file with sizes for reference alignment sequences (rnames). Must be tab delimited (chromosome\tsize) with one line per rname.',
);


#######################################################################
##########################   Consume Roles   ##########################
#######################################################################
with 
	"CLIPSeqTools::Role::Option::Library" => {
		-alias    => { validate_args => '_validate_args_for_library' },
		-excludes => 'validate_args',
	},
	"CLIPSeqTools::Role::Option::ReferenceLibrary" => {
		-alias    => { validate_args => '_validate_args_for_reference_library' },
		-excludes => 'validate_args',
	},
	"CLIPSeqTools::Role::Option::OutputPrefix" => {
		-alias    => { validate_args => '_validate_args_for_output_prefix' },
		-excludes => 'validate_args',
	};

	
#######################################################################
########################   Interface Methods   ########################
#######################################################################
sub validate_args {
	my ($self) = @_;
	
	$self->_validate_args_for_library;
	$self->_validate_args_for_reference_library;
	$self->_validate_args_for_output_prefix;
}

sub run {
	my ($self) = @_;
	
	warn "Starting analysis: libraries_overlap_stats\n";
	
	warn "Validating arguments\n" if $self->verbose;
	$self->validate_args();
	
	warn "Reading sizes for reference alignment sequences\n" if $self->verbose;
	my %rname_sizes = $self->read_rname_sizes;

	warn "Creating reads collection\n" if $self->verbose;
	my $reads_collection = $self->reads_collection;
	my @rnames = $reads_collection->rnames_for_all_strands;

	warn "Creating reference reads collection\n" if $self->verbose;
	my $r_reads_collection = $self->r_reads_collection;