use warnings;
use strict;
use Data::Dumper;
+use JSON::XS;
use Maasha::Common;
use Maasha::Filesys;
-use Maasha::NClist;
use Maasha::Align;
use vars qw( @ISA @EXPORT );
}
+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.
- # TODO find out what uses this and kill it!
-
# Parses a line with a KISS entry.
my ( $line, # KISS line to parse
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_index_old
+sub kiss_index
{
# Martin A. Hansen, December 2009.
# Creates a lookup index of a sorted KISS file.
- my ( $file, # path to KISS file
+ my ( $file, # path to KISS file
) = @_;
# Returns nothing.
}
-sub kiss_index
+sub kiss_index_offset
{
- # Martin A. Hansen, February 2010.
+ # Martin A. Hansen, December 2009.
- # Creates a NC list index of a sorted KISS file.
+ # Given a KISS index and a begin position,
+ # locate the offset closest to the begin position,
+ # and return this.
- my ( $file, # path to KISS file
+ my ( $index, # KISS index
+ $beg, # begin position
) = @_;
- # Returns nothing.
-
- my ( $fh, $line, @fields, $nc_list );
+ # Returns a number.
- $fh = Maasha::Filesys::file_read_open( $file );
+ my ( $bucket );
- while ( $line = <$fh> )
- {
- chomp $line;
+ Maasha::Common::error( qq(Negative begin position: "$beg") ) if $beg < 0;
- @fields = split "\t", $line;
+ $bucket = int( $beg / BUCKET_SIZE );
- if ( not defined $nc_list ) {
- $nc_list = [ [ @fields ] ];
- } else {
- Maasha::NClist::nc_list_add( $nc_list, [ @fields ], INDEX_END, INDEX );
- }
- }
+ $bucket = scalar @{ $index } if $bucket > scalar @{ $index };
- close $fh;
+ while ( $bucket >= 0 )
+ {
+ return $index->[ $bucket ]->[ OFFSET ] if defined $index->[ $bucket ];
- Maasha::NClist::nc_list_store( $nc_list, "$file.json" );
+ $bucket--;
+ }
}
-sub kiss_index_offset
+sub kiss_index_count
{
# Martin A. Hansen, December 2009.
- # Given a KISS index and a begin position,
- # locate the offset closest to the begin position,
+ # 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
+ my ( $index, # KISS index
+ $beg, # Begin position
+ $end, # End position
) = @_;
# Returns a number.
- my ( $bucket );
+ my ( $bucket_beg, $bucket_end, $count, $i );
Maasha::Common::error( qq(Negative begin position: "$beg") ) if $beg < 0;
- $bucket = int( $beg / BUCKET_SIZE );
+ $bucket_beg = int( $beg / BUCKET_SIZE );
+ $bucket_end = int( $end / BUCKET_SIZE );
- $bucket = scalar @{ $index } if $bucket > scalar @{ $index };
+ $bucket_end = scalar @{ $index } if $bucket_end > scalar @{ $index };
- while ( $bucket >= 0 )
- {
- return $index->[ $bucket ]->[ OFFSET ] if defined $index->[ $bucket ];
+ $count = 0;
- $bucket--;
+ for ( $i = $bucket_beg; $i <= $bucket_end; $i++ ) {
+ $count += $index->[ $i ]->[ COUNT ] if defined $index->[ $i ];
}
+
+ return $count;
}
-sub kiss_index_count
+sub kiss_index_count_nc
{
# Martin A. Hansen, December 2009.
# along with a beg/end interval, locate all entries
# in that interval and return those.
- my ( $index, # KISS index
+ my ( $file, # path to KISS file
+ $index, # KISS index
$beg, # interval begin
$end, # interval end
) = @_;
# Returns a list.
- my ( $features );
+ my ( $offset, $fh, $entry, @entries );
- $features = Maasha::NClist::nc_list_get_interval( $index, $beg, $end, INDEX_BEG, INDEX_END, INDEX );
+ # $offset = kiss_index_offset( $index, $beg );
- return wantarray ? @{ $features } : $features;
+ $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 );
+
+ 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;
}
# 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 );
- $index = Maasha::NClist::nc_list_retrieve( $path );
+ local $/ = undef;
+
+ $fh = Maasha::Filesys::file_read_open( $path );
+
+ $json = <$fh>;
+
+ close $fh;
+
+ $index = JSON::XS::decode_json( $json );
return wantarray ? @{ $index } : $index;
}
$entry->[ S_BEG ] = $record->{ 'S_BEG' };
$entry->[ S_END ] = $record->{ 'S_END' };
$entry->[ Q_ID ] = $record->{ 'Q_ID' } || ".";
- $entry->[ SCORE ] = $record->{ 'SCORE' } || $record->{ 'E_VAL' } || ".";
+ $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' } || ".";
+ $entry->[ BLOCK_BEGS ] = $record->{ 'BLOCK_BEGS' } || $record->{ 'Q_BEGS' } || ".";
$entry->[ BLOCK_LENS ] = $record->{ 'BLOCK_LENS' } || ".";
$entry->[ BLOCK_TYPE ] = $record->{ 'BLOCK_TYPE' } || ".";
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;
+}
projects / biopieces.git / blobdiff