# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> RUN SCRIPT <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-my $t0 = gettimeofday();
-
run_script( $script );
-my $t1 = gettimeofday();
-
-print STDERR "Program: $script" . ( " " x ( 25 - length( $script ) ) ) . sprintf( "Run time: %.4f\n", ( $t1 - $t0 ) );
-
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> SUBROUTINES <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
# Returns nothing.
- my ( $options, $in, $out );
+ my ( $t0, $t1, $options, $in, $out );
+
+ $t0 = gettimeofday();
$options = get_options( $script );
close $in if defined $in;
close $out;
+
+ $t1 = gettimeofday();
+
+ print STDERR "Program: $script" . ( " " x ( 25 - length( $script ) ) ) . sprintf( "Run time: %.4f\n", ( $t1 - $t0 ) ) if $options->{ 'verbose' };
}
@options = qw(
adaptor|a=s
mismatches|m=s
- no_remove|n
+ remove|r=s
offset|o=s
);
}
priority|p=f
use_score|u
visibility|v=s
- wiggle|w
- score|S
- log10|L
color|c=s
chunk_size|C=s
);
{
Maasha::Common::error( qq(Argument to --$opt must be octal or decimal - not "$options{ $opt }") );
}
+ elsif ( $opt eq "remove" and $script eq "remove_adaptor" and $options{ $opt } !~ /before|after|skip/ )
+ {
+ Maasha::Common::error( qq(Argument to --$opt must be before, after, or skip - not "$options{ $opt }") );
+ }
}
Maasha::Common::error( qq(no --database specified) ) if $script eq "create_blast_db" and not $options{ "database" };
{
# Martin A. Hansen, Juli 2008.
- # Read fixedStep wiggle format from stream or file.
+ # Read fixedstep wiggle format from stream or file.
my ( $in, # handle to in stream
$out, # handle to out stream
$record->{ "CHR" } = $1;
$record->{ "CHR_BEG" } = $2;
$record->{ "STEP" } = $3;
- $record->{ "VALS" } = join ",", @{ $entry };
+ $record->{ "VALS" } = join ";", @{ $entry };
}
put_record( $record, $out );
{
map { $_ = sprintf( "%.4f", Maasha::Calc::log10( $_ ) ) } @{ $block } if $options->{ "log10" };
- $record->{ "CHR" } = $chr;
- $record->{ "CHR_BEG" } = $beg_block;
- $record->{ "STEP" } = 1;
- $record->{ "VALS" } = join ";", @{ $block };
+ $record->{ "CHR" } = $chr;
+ $record->{ "CHR_BEG" } = $beg_block;
+ $record->{ "STEP" } = 1;
+ $record->{ "VALS" } = join ";", @{ $block };
+ $record->{ "REC_TYPE" } = "fixed_step";
put_record( $record, $out );
map { $_ = sprintf( "%.4f", Maasha::Calc::log10( $_ ) ) } @{ $block } if $options->{ "log10" };
- $record->{ "CHR" } = $chr;
- $record->{ "CHR_BEG" } = $beg_block;
- $record->{ "STEP" } = 1;
- $record->{ "VALS" } = join ";", @{ $block };
+ $record->{ "CHR" } = $chr;
+ $record->{ "CHR_BEG" } = $beg_block;
+ $record->{ "STEP" } = 1;
+ $record->{ "VALS" } = join ";", @{ $block };
+ $record->{ "REC_TYPE" } = "fixed_step";
put_record( $record, $out );
$phastcons_file = Maasha::Config::genome_phastcons( $options->{ "genome" } );
$phastcons_index = Maasha::Config::genome_phastcons_index( $options->{ "genome" } );
- $index = Maasha::UCSC::phastcons_index_retrieve( $phastcons_index );
+ $index = Maasha::UCSC::fixedstep_index_retrieve( $phastcons_index );
$fh_phastcons = Maasha::Common::read_open( $phastcons_file );
if ( defined $options->{ "chr" } and defined $options->{ "beg" } and ( defined $options->{ "end" } or defined $options->{ "len" } ) )
$options->{ "end" } = $options->{ "beg" } + $options->{ "len" } - 1;
}
- $scores = Maasha::UCSC::phastcons_index_lookup( $index, $fh_phastcons, $options->{ "chr" }, $options->{ "beg" }, $options->{ "end" }, $options->{ "flank" } );
+ $scores = Maasha::UCSC::fixedstep_index_lookup( $index, $fh_phastcons, $options->{ "chr" }, $options->{ "beg" }, $options->{ "end" }, $options->{ "flank" } );
$record->{ "CHR" } = $options->{ "chr" };
$record->{ "CHR_BEG" } = $options->{ "beg" } - $options->{ "flank" };
$vals = $record->{ 'VALS' };
- $vals =~ tr/,/\n/;
+ $vals =~ tr/;/\n/;
print $fh "$vals\n";
}
my ( $record, $adaptor, $seq, $adaptor_len, $seq_len, $offset, $max_match, $max_mismatch, $pos );
+ $options->{ "remove" } ||= "after";
+
$max_mismatch = $options->{ "mismatches" } || 0;
- $offset = $options->{ "offset" } || 15;
- $adaptor = $options->{ "adaptor" };
- $adaptor_len = length $adaptor;
- $adaptor = [ split //, uc $adaptor ];
+ $offset = $options->{ "offset" };
- $max_match = $adaptor_len - $max_mismatch;
+ if ( not defined $offset ) {
+ $offset = 0;
+ } else {
+ $offset--;
+ }
+
+ $adaptor = uc $options->{ "adaptor" };
+ $adaptor_len = length $adaptor;
while ( $record = get_record( $in ) )
{
if ( $record->{ "SEQ" } )
{
- $seq = $record->{ "SEQ" };
+ $seq = uc $record->{ "SEQ" };
$seq_len = length $seq;
- $seq = [ split //, uc $seq ];
- $pos = Maasha::Seq::find_adaptor( $adaptor, $seq, $adaptor_len, $seq_len, $offset, $max_match, $max_mismatch );
+ $pos = Maasha::Common::index_m( $seq, $adaptor, $seq_len, $adaptor_len, $offset, $max_mismatch );
$record->{ "ADAPTOR_POS" } = $pos;
- if ( $pos >= 0 and not $options->{ "no_remove" } )
+ if ( $pos >= 0 and $options->{ "remove" } ne "skip" )
{
- $record->{ "SEQ" } = substr $record->{ "SEQ" }, 0, $pos;
- $record->{ "SEQ_LEN" } = $pos;
+ if ( $options->{ "remove" } eq "after" )
+ {
+ $record->{ "SEQ" } = substr $record->{ "SEQ" }, 0, $pos;
+ $record->{ "SEQ_LEN" } = $pos;
+ }
+ else
+ {
+ $record->{ "SEQ" } = substr $record->{ "SEQ" }, $pos + $adaptor_len;
+ $record->{ "SEQ_LEN" } = length $record->{ "SEQ" };
+ }
}
put_record( $record, $out );
while ( $record = get_record( $in ) )
{
- $data_hash{ $record->{ $options->{ "key" } } }++ if $record->{ $options->{ "key" } };
+ $data_hash{ $record->{ $options->{ "key" } } }++ if defined $record->{ $options->{ "key" } };
put_record( $record, $out ) if not $options->{ "no_stream" };
}
while ( $record = get_record( $in ) )
{
- $data_hash{ $record->{ $options->{ "key" } } }++ if $record->{ $options->{ "key" } };
+ $data_hash{ $record->{ $options->{ "key" } } }++ if defined $record->{ $options->{ "key" } };
put_record( $record, $out ) if not $options->{ "no_stream" };
}
# Returns nothing.
- my ( $record, $file, $wib_file, $wig_file, $wib_dir, $fh_in, $fh_out, $i, $first, $format, $args, $type, $columns, $append, %fh_hash,
- $chr, $beg, $end, $block, $line, $max, $beg_block, $entry, $q_id, $clones, $vals );
+ my ( $record, $file, $wib_file, $wig_file, $wib_dir, $fh_out, $i, $first, $format, $type, $columns, $append, $vals );
$options->{ "short_label" } ||= $options->{ 'table' };
$options->{ "long_label" } ||= $options->{ 'table' };
$i = 0;
- if ( $options->{ 'wiggle' } )
+ $fh_out = Maasha::Common::write_open( $file );
+
+ while ( $record = get_record( $in ) )
{
- while ( $record = get_record( $in ) )
- {
- put_record( $record, $out ) if not $options->{ "no_stream" };
+ put_record( $record, $out ) if not $options->{ "no_stream" };
- $record->{ "CHR" } = $record->{ "S_ID" } if not defined $record->{ "CHR" };
- $record->{ "CHR_BEG" } = $record->{ "S_BEG" } if not defined $record->{ "CHR_BEG" };
- $record->{ "CHR_END" } = $record->{ "S_END" } if not defined $record->{ "CHR_END" };
+ if ( $record->{ "REC_TYPE" } eq "fixed_step" )
+ {
+ $vals = $record->{ "VALS" };
+ $vals =~ tr/;/\n/;
- $fh_hash{ $record->{ "CHR" } } = Maasha::Common::write_open( "$BP_TMP/$record->{ 'CHR' }" ) if not exists $fh_hash{ $record->{ "CHR" } };
+ print $fh_out "fixedStep chrom=$record->{ 'CHR' } start=$record->{ 'CHR_BEG' } step=$record->{ 'STEP' }\n";
+ print $fh_out "$vals\n";
- $fh_out = $fh_hash{ $record->{ "CHR" } };
-
- Maasha::UCSC::bed_put_entry( $record, $fh_out, 5 );
+ $format = "WIGGLE" if not $format;
}
-
- map { close $_ } keys %fh_hash;
-
- $fh_out = Maasha::Common::write_open( $file );
-
- foreach $chr ( sort keys %fh_hash )
+ elsif ( $record->{ "REC_TYPE" } eq "PSL" )
{
- Maasha::Common::run( "bedSort", "$BP_TMP/$chr $BP_TMP/$chr" );
-
- $fh_in = Maasha::Common::read_open( "$BP_TMP/$chr" );
-
- undef $block;
-
- while ( $entry = Maasha::UCSC::bed_get_entry( $fh_in, 5 ) )
- {
- $chr = $entry->{ 'CHR' };
- $beg = $entry->{ 'CHR_BEG' };
- $end = $entry->{ 'CHR_END' };
- $q_id = $entry->{ 'Q_ID' };
-
- if ( $options->{ "score" } ) {
- $clones = $entry->{ 'SCORE' };
- } elsif ( $q_id =~ /_(\d+)$/ ) {
- $clones = $1;
- } else {
- $clones = 1;
- }
-
- if ( $block )
- {
- if ( $beg > $max )
- {
- Maasha::UCSC::fixedstep_put_entry( $chr, $beg_block, $block, $fh_out, $options->{ "log10" } );
- undef $block;
- }
- else
- {
- for ( $i = $beg - $beg_block; $i < ( $beg - $beg_block ) + ( $end - $beg ); $i++ ) {
- $block->[ $i ] += $clones;
- }
-
- $max = Maasha::Calc::max( $max, $end );
- }
- }
-
- if ( not $block )
- {
- $beg_block = $beg;
- $max = $end;
-
- for ( $i = 0; $i < ( $end - $beg ); $i++ ) {
- $block->[ $i ] += $clones;
- }
- }
- }
-
- close $fh_in;
-
- Maasha::UCSC::fixedstep_put_entry( $chr, $beg_block, $block, $fh_out, $options->{ "log10" } );
+ Maasha::UCSC::psl_put_header( $fh_out ) if $first;
+ Maasha::UCSC::psl_put_entry( $record, $fh_out );
+
+ $first = 0;
- unlink "$BP_TMP/$chr";
+ $format = "PSL" if not $format;
}
-
- close $fh_out;
-
- $format = "WIGGLE";
- }
- else
- {
- $fh_out = Maasha::Common::write_open( $file );
-
- while ( $record = get_record( $in ) )
+ elsif ( $record->{ "REC_TYPE" } eq "BED" and $record->{ "SEC_STRUCT" } )
{
- put_record( $record, $out ) if not $options->{ "no_stream" };
+ # chrom chromStart chromEnd name score strand size secStr conf
- if ( $record->{ "REC_TYPE" } eq "fixed_step" )
- {
- $vals = $record->{ "VALS" };
- $vals =~ tr/,/\n/;
+ print $fh_out join ( "\t",
+ $record->{ "CHR" },
+ $record->{ "CHR_BEG" },
+ $record->{ "CHR_END" } + 1,
+ $record->{ "Q_ID" },
+ $record->{ "SCORE" },
+ $record->{ "STRAND" },
+ $record->{ "SIZE" },
+ $record->{ "SEC_STRUCT" },
+ $record->{ "CONF" },
+ ), "\n";
- print $fh_out "fixedStep chrom=$record->{ 'CHR' } start=$record->{ 'CHR_BEG' } step=$record->{ 'STEP' }\n";
- print $fh_out "$vals\n";
+ $format = "BED_SS" if not $format;
+ }
+ elsif ( $record->{ "REC_TYPE" } eq "BED" )
+ {
+ Maasha::UCSC::bed_put_entry( $record, $fh_out, $record->{ "BED_COLS" } );
- $format = "WIGGLE" if not $format;
- }
- elsif ( $record->{ "REC_TYPE" } eq "PSL" )
- {
- Maasha::UCSC::psl_put_header( $fh_out ) if $first;
- Maasha::UCSC::psl_put_entry( $record, $fh_out );
-
- $first = 0;
+ $format = "BED" if not $format;
+ $columns = $record->{ "BED_COLS" } if not $columns;
+ }
+ elsif ( $record->{ "REC_TYPE" } eq "PATSCAN" and $record->{ "CHR" } )
+ {
+ Maasha::UCSC::bed_put_entry( $record, $fh_out, 6 );
- $format = "PSL" if not $format;
- }
- elsif ( $record->{ "REC_TYPE" } eq "BED" and $record->{ "SEC_STRUCT" } )
- {
- # chrom chromStart chromEnd name score strand size secStr conf
-
- print $fh_out join ( "\t",
- $record->{ "CHR" },
- $record->{ "CHR_BEG" },
- $record->{ "CHR_END" } + 1,
- $record->{ "Q_ID" },
- $record->{ "SCORE" },
- $record->{ "STRAND" },
- $record->{ "SIZE" },
- $record->{ "SEC_STRUCT" },
- $record->{ "CONF" },
- ), "\n";
-
- $format = "BED_SS" if not $format;
- }
- elsif ( $record->{ "REC_TYPE" } eq "BED" )
- {
- Maasha::UCSC::bed_put_entry( $record, $fh_out, $record->{ "BED_COLS" } );
+ $format = "BED" if not $format;
+ $columns = 6 if not $columns;
+ }
+ elsif ( $record->{ "REC_TYPE" } eq "BLAST" and $record->{ "S_ID" } =~ /^chr/ )
+ {
+ $record->{ "CHR" } = $record->{ "S_ID" };
+ $record->{ "CHR_BEG" } = $record->{ "S_BEG" };
+ $record->{ "CHR_END" } = $record->{ "S_END" };
+ $record->{ "SCORE" } = $record->{ "BIT_SCORE" } * 1000;
- $format = "BED" if not $format;
- $columns = $record->{ "BED_COLS" } if not $columns;
- }
- elsif ( $record->{ "REC_TYPE" } eq "PATSCAN" and $record->{ "CHR" } )
- {
- Maasha::UCSC::bed_put_entry( $record, $fh_out, 6 );
+ $format = "BED" if not $format;
+ $columns = 6 if not $columns;
- $format = "BED" if not $format;
- $columns = 6 if not $columns;
- }
- elsif ( $record->{ "REC_TYPE" } eq "BLAST" and $record->{ "S_ID" } =~ /^chr/ )
- {
- $record->{ "CHR" } = $record->{ "S_ID" };
- $record->{ "CHR_BEG" } = $record->{ "S_BEG" };
- $record->{ "CHR_END" } = $record->{ "S_END" };
- $record->{ "SCORE" } = $record->{ "BIT_SCORE" } * 1000;
+ Maasha::UCSC::bed_put_entry( $record, $fh_out );
+ }
+ elsif ( $record->{ "REC_TYPE" } eq "VMATCH" and $record->{ "S_ID" } =~ /^chr/i )
+ {
+ $record->{ "CHR" } = $record->{ "S_ID" };
+ $record->{ "CHR_BEG" } = $record->{ "S_BEG" };
+ $record->{ "CHR_END" } = $record->{ "S_END" };
+ $record->{ "SCORE" } = $record->{ "SCORE" } || 999;
+ $record->{ "SCORE" } = int( $record->{ "SCORE" } );
- $format = "BED" if not $format;
- $columns = 6 if not $columns;
+ $format = "BED" if not $format;
+ $columns = 6 if not $columns;
- Maasha::UCSC::bed_put_entry( $record, $fh_out );
- }
- elsif ( $record->{ "REC_TYPE" } eq "VMATCH" and $record->{ "S_ID" } =~ /^chr/i )
- {
- $record->{ "CHR" } = $record->{ "S_ID" };
- $record->{ "CHR_BEG" } = $record->{ "S_BEG" };
- $record->{ "CHR_END" } = $record->{ "S_END" };
- $record->{ "SCORE" } = $record->{ "SCORE" } || 999;
- $record->{ "SCORE" } = int( $record->{ "SCORE" } );
+ Maasha::UCSC::bed_put_entry( $record, $fh_out, 6 );
+ }
- $format = "BED" if not $format;
- $columns = 6 if not $columns;
+ if ( $i == $options->{ "chunk_size" } )
+ {
+ close $fh_out;
- Maasha::UCSC::bed_put_entry( $record, $fh_out, 6 );
+ if ( $format eq "BED" ) {
+ Maasha::UCSC::bed_upload_to_ucsc( $BP_TMP, $file, $options, $append );
+ } elsif ( $format eq "PSL" ) {
+ Maasha::UCSC::psl_upload_to_ucsc( $file, $options, $append );
}
- if ( $i == $options->{ "chunk_size" } )
- {
- close $fh_out;
-
- if ( $format eq "BED" ) {
- Maasha::UCSC::bed_upload_to_ucsc( $BP_TMP, $file, $options, $append );
- } elsif ( $format eq "PSL" ) {
- Maasha::UCSC::psl_upload_to_ucsc( $file, $options, $append );
- }
-
- unlink $file;
-
- $first = 1;
+ unlink $file;
- $append = 1;
+ $first = 1;
- $fh_out = Maasha::Common::write_open( $file );
- }
+ $append = 1;
- $i++;
+ $fh_out = Maasha::Common::write_open( $file );
}
+
+ $i++;
}
close $fh_out;
{
# Martin A. Hansen, July 2008.
- # Cleans out any unused temporary files and direcotries in BP_TMP.
+ # Cleans out any unused temporary files and directories in BP_TMP.
# Returns nothing.
elsif ( $pid == $curr_pid )
{
# print STDERR "Removing current dir: $dir\n";
- Maasha::Common::dir_remove( $dir );
+ # Maasha::Common::dir_remove( $dir );
}
}
}
projects / biopieces.git / blobdiff