use warnings;
use strict;
use Data::Dumper;
+use JSON::XS;
use Maasha::Common;
use Maasha::Filesys;
use Maasha::Align;
+
use vars qw( @ISA @EXPORT );
@ISA = qw( Exporter );
BUCKET_SIZE => 100,
COUNT => 0,
OFFSET => 1,
+
+ INDEX_BEG => 1,
+ INDEX_END => 2,
+ INDEX => 12,
};
# Returns a hashref.
- my ( $line, @fields, %entry );
+ my ( $line, @fields );
while ( $line = <$fh> )
{
Maasha::Common::error( qq(BAD kiss entry: $line) ) if not @fields == 12;
- $entry{ 'S_ID' } = $fields[ S_ID ];
- $entry{ 'S_BEG' } = $fields[ S_BEG ];
- $entry{ 'S_END' } = $fields[ S_END ];
- $entry{ 'Q_ID' } = $fields[ Q_ID ];
- $entry{ 'SCORE' } = $fields[ SCORE ];
- $entry{ 'STRAND' } = $fields[ STRAND ];
- $entry{ 'HITS' } = $fields[ HITS ];
- $entry{ 'ALIGN' } = $fields[ ALIGN ];
- $entry{ 'BLOCK_COUNT' } = $fields[ BLOCK_COUNT ];
- $entry{ 'BLOCK_BEGS' } = $fields[ BLOCK_BEGS ];
- $entry{ 'BLOCK_LENS' } = $fields[ BLOCK_LENS ];
- $entry{ 'BLOCK_TYPE' } = $fields[ BLOCK_TYPE ];
-
- return wantarray ? %entry : \%entry;
+ return wantarray ? @fields : \@fields;
}
}
+sub kiss_retrieve
+{
+ # Martin A. Hansen, February 2010.
+
+ # Retrieves KISS entries from a given sorted KISS file
+ # within an optional interval.
+
+ my ( $file, # path to KISS file
+ $beg, # interval begin - OPTIONAL
+ $end, # interval end - OPTIONAL
+ ) = @_;
+
+ # Returns a list.
+
+ my ( $fh, $entry, @entries );
+
+ $beg ||= 0;
+ $end ||= 999999999;
+
+ $fh = Maasha::Filesys::file_read_open( $file );
+
+ while ( $entry = kiss_entry_get( $fh ) )
+ {
+ last if $entry->[ S_BEG ] > $end;
+ push @entries, $entry if $entry->[ S_END ] > $beg;
+ }
+
+ close $fh;
+
+ return wantarray ? @entries : \@entries;
+}
+
+
sub kiss_entry_parse
{
# Martin A. Hansen, December 2009.
Maasha::Common::error( qq(BAD kiss entry: $line) ) if not @fields == 12;
- $entry{ 'S_ID' } = $fields[ S_ID ];
- $entry{ 'S_BEG' } = $fields[ S_BEG ];
- $entry{ 'S_END' } = $fields[ S_END ];
- $entry{ 'Q_ID' } = $fields[ Q_ID ];
- $entry{ 'SCORE' } = $fields[ SCORE ];
- $entry{ 'STRAND' } = $fields[ STRAND ];
- $entry{ 'HITS' } = $fields[ HITS ];
- $entry{ 'ALIGN' } = $fields[ ALIGN ];
- $entry{ 'BLOCK_COUNT' } = $fields[ BLOCK_COUNT ];
- $entry{ 'BLOCK_BEGS' } = $fields[ BLOCK_BEGS ];
- $entry{ 'BLOCK_LENS' } = $fields[ BLOCK_LENS ];
- $entry{ 'BLOCK_TYPE' } = $fields[ BLOCK_TYPE ];
-
- return wantarray ? %entry : \%entry;
+ return wantarray ? @fields : \@fields;
}
# Returns nothing.
- my ( @fields );
+ Maasha::Common::error( qq(BAD kiss entry) ) if not scalar @{ $entry } == 12;
- if ( defined $entry->{ 'S_ID' } and
- defined $entry->{ 'S_BEG' } and
- defined $entry->{ 'S_END' }
+ if ( defined $entry->[ S_ID ] and
+ defined $entry->[ S_BEG ] and
+ defined $entry->[ S_END ]
)
{
- Maasha::Common::error( qq(Bad S_BEG value: $entry->{ 'S_BEG' } < 0 ) ) if $entry->{ 'S_BEG' } < 0;
- Maasha::Common::error( qq(Bad S_END value: $entry->{ 'S_END' } < $entry->{ 'S_BEG' }) ) if $entry->{ 'S_END' } < $entry->{ 'S_BEG' };
+ Maasha::Common::error( qq(Bad S_BEG value: $entry->[ S_BEG ] < 0 ) ) if $entry->[ S_BEG ] < 0;
+ Maasha::Common::error( qq(Bad S_END value: $entry->[ S_END ] < $entry->[ S_BEG ] ) ) if $entry->[ S_END ] < $entry->[ S_BEG ];
$fh ||= \*STDOUT;
- $fields[ S_ID ] = $entry->{ 'S_ID' };
- $fields[ S_BEG ] = $entry->{ 'S_BEG' };
- $fields[ S_END ] = $entry->{ 'S_END' };
- $fields[ Q_ID ] = $entry->{ 'Q_ID' } || ".";
- $fields[ SCORE ] = $entry->{ 'SCORE' } || ".";
- $fields[ STRAND ] = $entry->{ 'STRAND' } || ".";
- $fields[ HITS ] = $entry->{ 'HITS' } || ".";
- $fields[ ALIGN ] = $entry->{ 'ALIGN' } || ".";
- $fields[ BLOCK_COUNT ] = $entry->{ 'BLOCK_COUNT' } || ".";
- $fields[ BLOCK_BEGS ] = $entry->{ 'BLOCK_BEGS' } || ".";
- $fields[ BLOCK_LENS ] = $entry->{ 'BLOCK_LENS' } || ".";
- $fields[ BLOCK_TYPE ] = $entry->{ 'BLOCK_TYPE' } || ".";
-
- print $fh join( "\t", @fields ), "\n";
+ $entry->[ Q_ID ] = "." if not defined $entry->[ Q_ID ];
+ $entry->[ SCORE ] = "." if not defined $entry->[ SCORE ];
+ $entry->[ STRAND ] = "." if not defined $entry->[ STRAND ];
+ $entry->[ HITS ] = "." if not defined $entry->[ HITS ];
+ $entry->[ ALIGN ] = "." if not defined $entry->[ ALIGN ];
+ $entry->[ BLOCK_COUNT ] = "." if not defined $entry->[ BLOCK_COUNT ];
+ $entry->[ BLOCK_BEGS ] = "." if not defined $entry->[ BLOCK_BEGS ];
+ $entry->[ BLOCK_LENS ] = "." if not defined $entry->[ BLOCK_LENS ];
+ $entry->[ BLOCK_TYPE ] = "." if not defined $entry->[ BLOCK_TYPE ];
+
+ print $fh join( "\t", @{ $entry } ), "\n";
}
}
# Returns nothing.
- `sort -k 2,2n -k 3,3n $file > $file.sort`;
+ `sort -k 2,2n -k 3,3nr $file > $file.sort`;
rename "$file.sort", $file;
}
# Creates a lookup index of a sorted KISS file.
- my ( $file, # path to KISS file
+ my ( $file, # path to KISS file
) = @_;
# Returns nothing.
Maasha::Common::error( qq(Negative begin position: "$beg") ) if $beg < 0;
- $bucket_beg = int( $beg / BUCKET_SIZE );
- $bucket_end = int( $end / BUCKET_SIZE );
+ $bucket_beg = int( $beg / BUCKET_SIZE );
+ $bucket_end = int( $end / BUCKET_SIZE );
$bucket_end = scalar @{ $index } if $bucket_end > scalar @{ $index };
}
+sub kiss_index_count_nc
+{
+ # Martin A. Hansen, December 2009.
+
+ # Given a KISS index and a begin/end interval
+ # sum the number of counts in that interval,
+ # and return this.
+
+ my ( $index, # KISS index
+ $beg, # Begin position
+ $end, # End position
+ ) = @_;
+
+ # Returns a number.
+
+ my ( $count );
+
+ Maasha::Common::error( qq(Negative begin position: "$beg") ) if $beg < 0;
+
+ $count = Maasha::NClist::nc_list_count_interval( $index, $beg, $end, INDEX_BEG, INDEX_END, INDEX );
+
+ return $count;
+}
+
+
sub kiss_index_get_entries
{
# Martin A. Hansen, November 2009.
my ( $offset, $fh, $entry, @entries );
+ # $offset = kiss_index_offset( $index, $beg );
+
+ $fh = Maasha::Filesys::file_read_open( $file );
+
+ # sysseek( $fh, $offset, 0 );
+
+ while ( $entry = Maasha::KISS::kiss_entry_get( $fh ) )
+ {
+ push @entries, $entry if $entry->[ S_END ] > $beg;
+
+ last if $entry->[ S_BEG ] > $end;
+ }
+
+ close $fh;
+
+ return wantarray ? @entries : \@entries;
+}
+
+
+sub kiss_index_get_entries_OLD
+{
+ # Martin A. Hansen, November 2009.
+
+ # Given a path to a KISS file and a KISS index
+ # along with a beg/end interval, locate all entries
+ # in that interval and return those.
+
+ my ( $file, # path to KISS file
+ $index, # KISS index
+ $beg, # interval begin
+ $end, # interval end
+ ) = @_;
+
+ # Returns a list.
+
+ my ( $offset, $fh, $entry, @entries );
+
$offset = kiss_index_offset( $index, $beg );
$fh = Maasha::Filesys::file_read_open( $file );
while ( $entry = Maasha::KISS::kiss_entry_get( $fh ) )
{
- push @entries, $entry if $entry->{ 'S_END' } > $beg;
-
- last if $entry->{ 'S_BEG' } > $end;
+ push @entries, $entry if $entry->[ S_END ] > $beg;
+
+ last if $entry->[ S_BEG ] > $end;
}
close $fh;
my ( $entry, %lookup, $pos, $overlap, @entries );
while ( $entry = kiss_entry_get( $fh2 ) ) {
- map { $lookup{ $_ } = 1 } ( $entry->{ 'S_BEG' } .. $entry->{ 'S_END' } );
+ map { $lookup{ $_ } = 1 } ( $entry->[ S_BEG ] .. $entry->[ S_END ] );
}
while ( $entry = kiss_entry_get( $fh1 ) )
{
$overlap = 0;
- foreach $pos ( $entry->{ 'S_BEG' } .. $entry->{ 'S_END' } )
+ foreach $pos ( $entry->[ S_BEG ] .. $entry->[ S_END ] )
{
if ( exists $lookup{ $pos } )
{
# Returns nothing.
- Maasha::Filesys::file_store( $path, $index );
+ my ( $fh, $json );
+
+ $json = JSON::XS::encode_json( $index );
+
+ $fh = Maasha::Filesys::file_write_open( $path );
+
+ print $fh $json;
+
+ close $fh;
}
# Returns a data structure.
- my ( $index );
+ my ( $fh, $json, $index );
+
+ local $/ = undef;
- $index = Maasha::Filesys::file_retrieve( $path );
+ $fh = Maasha::Filesys::file_read_open( $path );
+
+ $json = <$fh>;
+
+ close $fh;
+
+ $index = JSON::XS::decode_json( $json );
return wantarray ? @{ $index } : $index;
}
# Martin A. Hansen, November 2009.
# Converts a KISS entry to a Biopiece record.
- # TODO: Consistency checking
my ( $entry, # KISS entry
) = @_;
# Returns a hashref
- return wantarray ? %{ $entry } : $entry;
+ my ( %record );
+
+ Maasha::Common::error( qq(BAD kiss entry) ) if not scalar @{ $entry } == 12;
+
+ $record{ 'S_ID' } = $entry->[ S_ID ];
+ $record{ 'S_BEG' } = $entry->[ S_BEG ];
+ $record{ 'S_END' } = $entry->[ S_END ];
+ $record{ 'Q_ID' } = $entry->[ Q_ID ];
+ $record{ 'SCORE' } = $entry->[ SCORE ];
+ $record{ 'STRAND' } = $entry->[ STRAND ];
+ $record{ 'HITS' } = $entry->[ HITS ];
+ $record{ 'ALIGN' } = $entry->[ ALIGN ];
+ $record{ 'BLOCK_COUNT' } = $entry->[ BLOCK_COUNT ];
+ $record{ 'BLOCK_BEGS' } = $entry->[ BLOCK_BEGS ];
+ $record{ 'BLOCK_LENS' } = $entry->[ BLOCK_LENS ];
+ $record{ 'BLOCK_TYPE' } = $entry->[ BLOCK_TYPE ];
+
+ return wantarray ? %record : \%record;
}
# Returns a hashref
+ my ( $entry );
+
if ( not defined $record->{ 'S_ID' } and
not defined $record->{ 'S_BEG' } and
not defined $record->{ 'S_END' } )
return undef;
}
- $record->{ 'SCORE' } ||= $record->{ 'E_VAL' } || ".";
- $record->{ 'HITS' } ||= ".";
- $record->{ 'BLOCK_COUNT' } ||= ".";
- $record->{ 'BLOCK_BEGS' } ||= ".";
- $record->{ 'BLOCK_LENS' } ||= ".";
- $record->{ 'BLOCK_TYPE' } ||= ".";
- $record->{ 'ALIGN' } ||= $record->{ 'DESCRIPTOR' } || ".";
-
- return wantarray ? %{ $record } : $record;
+ $entry->[ S_ID ] = $record->{ 'S_ID' };
+ $entry->[ S_BEG ] = $record->{ 'S_BEG' };
+ $entry->[ S_END ] = $record->{ 'S_END' };
+ $entry->[ Q_ID ] = $record->{ 'Q_ID' } || ".";
+ $entry->[ SCORE ] = $record->{ 'SCORE' } || $record->{ 'BIT_SCORE' } || $record->{ 'ID' } || ".";
+ $entry->[ STRAND ] = $record->{ 'STRAND' } || ".";
+ $entry->[ HITS ] = $record->{ 'HITS' } || ".";
+ $entry->[ ALIGN ] = $record->{ 'ALIGN' } || $record->{ 'DESCRIPTOR' } || ".";
+ $entry->[ BLOCK_COUNT ] = $record->{ 'BLOCK_COUNT' } || ".";
+ $entry->[ BLOCK_BEGS ] = $record->{ 'BLOCK_BEGS' } || $record->{ 'Q_BEGS' } || ".";
+ $entry->[ BLOCK_LENS ] = $record->{ 'BLOCK_LENS' } || ".";
+ $entry->[ BLOCK_TYPE ] = $record->{ 'BLOCK_TYPE' } || ".";
+
+ return wantarray ? @{ $entry } : $entry;
}
1;
+__END__
+
+sub kiss_index_nc
+{
+ # Martin A. Hansen, February 2010.
+
+ # Creates a NC list index of a sorted KISS file.
+
+ my ( $file, # path to KISS file
+ ) = @_;
+
+ # Returns nothing.
+
+ my ( $fh, $line, @fields, $nc_list );
+
+ $fh = Maasha::Filesys::file_read_open( $file );
+
+ while ( $line = <$fh> )
+ {
+ chomp $line;
+
+ @fields = split "\t", $line;
+
+ if ( not defined $nc_list ) {
+ $nc_list = [ [ @fields ] ];
+ } else {
+ Maasha::NClist::nc_list_add( $nc_list, [ @fields ], INDEX_END, INDEX );
+ }
+ }
+
+ close $fh;
+
+ Maasha::NClist::nc_list_store( $nc_list, "$file.json" );
+}
+
+
+sub kiss_index_get_entries_nc
+{
+ # Martin A. Hansen, November 2009.
+
+ # Given a path to a KISS file and a KISS index
+ # along with a beg/end interval, locate all entries
+ # in that interval and return those.
+
+ my ( $index, # KISS index
+ $beg, # interval begin
+ $end, # interval end
+ ) = @_;
+
+ # Returns a list.
+
+ my ( $features );
+
+ $features = Maasha::NClist::nc_list_get_interval( $index, $beg, $end, INDEX_BEG, INDEX_END, INDEX );
+
+ return wantarray ? @{ $features } : $features;
+}
+
+
+sub kiss_index_retrieve_nc
+{
+ # Martin A. Hansen, November 2009.
+
+ # Retrieves a KISS index from a file.
+
+ my ( $path, # Path to KISS index
+ ) = @_;
+
+ # Returns a data structure.
+
+ my ( $index );
+
+ $index = Maasha::NClist::nc_list_retrieve( $path );
+
+ return wantarray ? @{ $index } : $index;
+}