added read_fixedstep

git-svn-id: http://biopieces.googlecode.com/svn/trunk@144 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_bin/read_fixedstep b/bp_bin/read_fixedstep
new file mode 100755 (executable)
index 0000000..4cd1d44
--- /dev/null
+++ b/bp_bin/read_fixedstep
@@ -0,0 +1,6 @@
+#!/usr/bin/env perl
+
+use warnings;
+use strict;
+
+use Maasha::Biopieces;

diff --git a/code_perl/Maasha/Biopieces.pm b/code_perl/Maasha/Biopieces.pm
index 8c9358b543123d1facab6b11e1f0abcc7be365b7..aac2215d36cee9882e470fb0321da83bc3d197b4 100644 (file)
--- a/code_perl/Maasha/Biopieces.pm
+++ b/code_perl/Maasha/Biopieces.pm
@@ -164,93 +164,94 @@ sub run_script
     $in  = read_stream( $options->{ "stream_in" } );
     $out = write_stream( $options->{ "stream_out" } );
 
-    if    ( $script eq "print_usage" )              { script_print_usage( $in, $out, $options ) }
-    elsif ( $script eq "list_biopieces" )           { script_list_biopieces( $in, $out, $options ) }
-    elsif ( $script eq "list_genomes" )             { script_list_genomes( $in, $out, $options ) }
-    elsif ( $script eq "read_fasta" )               { script_read_fasta( $in, $out, $options ) }
-    elsif ( $script eq "read_tab" )                 { script_read_tab( $in, $out, $options ) }
-    elsif ( $script eq "read_psl" )                 { script_read_psl( $in, $out, $options ) }
-    elsif ( $script eq "read_bed" )                 { script_read_bed( $in, $out, $options ) }
-    elsif ( $script eq "read_blast_tab" )           { script_read_blast_tab( $in, $out, $options ) }
-    elsif ( $script eq "read_embl" )                { script_read_embl( $in, $out, $options ) }
-    elsif ( $script eq "read_stockholm" )           { script_read_stockholm( $in, $out, $options ) }
-    elsif ( $script eq "read_phastcons" )           { script_read_phastcons( $in, $out, $options ) }
-    elsif ( $script eq "read_soft" )                { script_read_soft( $in, $out, $options ) }
-    elsif ( $script eq "read_gff" )                 { script_read_gff( $in, $out, $options ) }
-    elsif ( $script eq "read_2bit" )                { script_read_2bit( $in, $out, $options ) }
-    elsif ( $script eq "read_solexa" )              { script_read_solexa( $in, $out, $options ) }
-    elsif ( $script eq "read_solid" )               { script_read_solid( $in, $out, $options ) }
-    elsif ( $script eq "read_mysql" )               { script_read_mysql( $in, $out, $options ) }
-    elsif ( $script eq "format_genome" )            { script_format_genome( $in, $out, $options ) }
-    elsif ( $script eq "length_seq" )               { script_length_seq( $in, $out, $options ) }
-    elsif ( $script eq "uppercase_seq" )            { script_uppercase_seq( $in, $out, $options ) }
-    elsif ( $script eq "shuffle_seq" )              { script_shuffle_seq( $in, $out, $options ) }
-    elsif ( $script eq "analyze_seq" )              { script_analyze_seq( $in, $out, $options ) }
-    elsif ( $script eq "analyze_tags" )             { script_analyze_tags( $in, $out, $options ) }
-    elsif ( $script eq "complexity_seq" )           { script_complexity_seq( $in, $out, $options ) }
-    elsif ( $script eq "oligo_freq" )               { script_oligo_freq( $in, $out, $options ) }
-    elsif ( $script eq "create_weight_matrix" )     { script_create_weight_matrix( $in, $out, $options ) }
-    elsif ( $script eq "calc_bit_scores" )          { script_calc_bit_scores( $in, $out, $options ) }
-    elsif ( $script eq "reverse_seq" )              { script_reverse_seq( $in, $out, $options ) }
-    elsif ( $script eq "complement_seq" )           { script_complement_seq( $in, $out, $options ) }
-    elsif ( $script eq "remove_indels" )            { script_remove_indels( $in, $out, $options ) }
-    elsif ( $script eq "transliterate_seq" )        { script_transliterate_seq( $in, $out, $options ) }
-    elsif ( $script eq "transliterate_vals" )       { script_transliterate_vals( $in, $out, $options ) }
-    elsif ( $script eq "translate_seq" )            { script_translate_seq( $in, $out, $options ) }
-    elsif ( $script eq "extract_seq" )              { script_extract_seq( $in, $out, $options ) }
-    elsif ( $script eq "get_genome_seq" )           { script_get_genome_seq( $in, $out, $options ) }
-    elsif ( $script eq "get_genome_align" )         { script_get_genome_align( $in, $out, $options ) }
-    elsif ( $script eq "get_genome_phastcons" )     { script_get_genome_phastcons( $in, $out, $options ) }
-    elsif ( $script eq "fold_seq" )                 { script_fold_seq( $in, $out, $options ) }
-    elsif ( $script eq "split_seq" )                { script_split_seq( $in, $out, $options ) }
-    elsif ( $script eq "split_bed" )                { script_split_bed( $in, $out, $options ) }
-    elsif ( $script eq "align_seq" )                { script_align_seq( $in, $out, $options ) }
-    elsif ( $script eq "tile_seq" )                 { script_tile_seq( $in, $out, $options ) }
-    elsif ( $script eq "invert_align" )             { script_invert_align( $in, $out, $options ) }
-    elsif ( $script eq "patscan_seq" )              { script_patscan_seq( $in, $out, $options ) }
-    elsif ( $script eq "create_blast_db" )          { script_create_blast_db( $in, $out, $options ) }
-    elsif ( $script eq "blast_seq" )                { script_blast_seq( $in, $out, $options ) }
-    elsif ( $script eq "blat_seq" )                 { script_blat_seq( $in, $out, $options ) }
-    elsif ( $script eq "match_seq" )                { script_match_seq( $in, $out, $options ) }
-    elsif ( $script eq "create_vmatch_index" )      { script_create_vmatch_index( $in, $out, $options ) }
-    elsif ( $script eq "vmatch_seq" )               { script_vmatch_seq( $in, $out, $options ) }
-    elsif ( $script eq "write_fasta" )              { script_write_fasta( $in, $out, $options, $options ) }
-    elsif ( $script eq "write_align" )              { script_write_align( $in, $out, $options ) }
-    elsif ( $script eq "write_blast" )              { script_write_blast( $in, $out, $options ) }
-    elsif ( $script eq "write_tab" )                { script_write_tab( $in, $out, $options ) }
-    elsif ( $script eq "write_bed" )                { script_write_bed( $in, $out, $options ) }
-    elsif ( $script eq "write_psl" )                { script_write_psl( $in, $out, $options ) }
-    elsif ( $script eq "write_2bit" )               { script_write_2bit( $in, $out, $options, $options ) }
-    elsif ( $script eq "write_solid" )              { script_write_solid( $in, $out, $options, $options ) }
-    elsif ( $script eq "head_records" )             { script_head_records( $in, $out, $options ) }
-    elsif ( $script eq "remove_keys" )              { script_remove_keys( $in, $out, $options ) }
-    elsif ( $script eq "rename_keys" )              { script_rename_keys( $in, $out, $options ) }
-    elsif ( $script eq "uniq_vals" )                { script_uniq_vals( $in, $out, $options ) }
-    elsif ( $script eq "merge_vals" )               { script_merge_vals( $in, $out, $options ) }
-    elsif ( $script eq "grab" )                     { script_grab( $in, $out, $options ) }
-    elsif ( $script eq "compute" )                  { script_compute( $in, $out, $options ) }
-    elsif ( $script eq "flip_tab" )                 { script_flip_tab( $in, $out, $options ) }
-    elsif ( $script eq "add_ident" )                { script_add_ident( $in, $out, $options ) }
-    elsif ( $script eq "count_records" )            { script_count_records( $in, $out, $options ) }
-    elsif ( $script eq "random_records" )           { script_random_records( $in, $out, $options ) }
-    elsif ( $script eq "sort_records" )             { script_sort_records( $in, $out, $options ) }
-    elsif ( $script eq "count_vals" )               { script_count_vals( $in, $out, $options ) }
-    elsif ( $script eq "plot_histogram" )           { script_plot_histogram( $in, $out, $options ) }
-    elsif ( $script eq "plot_lendist" )             { script_plot_lendist( $in, $out, $options ) }
-    elsif ( $script eq "plot_chrdist" )             { script_plot_chrdist( $in, $out, $options ) }
-    elsif ( $script eq "plot_karyogram" )           { script_plot_karyogram( $in, $out, $options ) }
-    elsif ( $script eq "plot_matches" )             { script_plot_matches( $in, $out, $options ) }
-    elsif ( $script eq "plot_seqlogo" )             { script_plot_seqlogo( $in, $out, $options ) }
-    elsif ( $script eq "plot_phastcons_profiles" )  { script_plot_phastcons_profiles( $in, $out, $options ) }
-    elsif ( $script eq "analyze_bed" )              { script_analyze_bed( $in, $out, $options ) }
-    elsif ( $script eq "analyze_vals" )             { script_analyze_vals( $in, $out, $options ) }
-    elsif ( $script eq "length_vals" )              { script_length_vals( $in, $out, $options ) }
-    elsif ( $script eq "sum_vals" )                 { script_sum_vals( $in, $out, $options ) }
-    elsif ( $script eq "mean_vals" )                { script_mean_vals( $in, $out, $options ) }
-    elsif ( $script eq "median_vals" )              { script_median_vals( $in, $out, $options ) }
-    elsif ( $script eq "max_vals" )                 { script_max_vals( $in, $out, $options ) }
-    elsif ( $script eq "min_vals" )                 { script_min_vals( $in, $out, $options ) }
-    elsif ( $script eq "upload_to_ucsc" )           { script_upload_to_ucsc( $in, $out, $options ) }
+    if    ( $script eq "print_usage" )              { script_print_usage(               $in, $out, $options ) }
+    elsif ( $script eq "list_biopieces" )           { script_list_biopieces(            $in, $out, $options ) }
+    elsif ( $script eq "list_genomes" )             { script_list_genomes(              $in, $out, $options ) }
+    elsif ( $script eq "read_fasta" )               { script_read_fasta(                $in, $out, $options ) }
+    elsif ( $script eq "read_tab" )                 { script_read_tab(                  $in, $out, $options ) }
+    elsif ( $script eq "read_psl" )                 { script_read_psl(                  $in, $out, $options ) }
+    elsif ( $script eq "read_bed" )                 { script_read_bed(                  $in, $out, $options ) }
+    elsif ( $script eq "read_fixedstep" )           { script_read_fixedstep(            $in, $out, $options ) }
+    elsif ( $script eq "read_blast_tab" )           { script_read_blast_tab(            $in, $out, $options ) }
+    elsif ( $script eq "read_embl" )                { script_read_embl(                 $in, $out, $options ) }
+    elsif ( $script eq "read_stockholm" )           { script_read_stockholm(            $in, $out, $options ) }
+    elsif ( $script eq "read_phastcons" )           { script_read_phastcons(            $in, $out, $options ) }
+    elsif ( $script eq "read_soft" )                { script_read_soft(                 $in, $out, $options ) }
+    elsif ( $script eq "read_gff" )                 { script_read_gff(                  $in, $out, $options ) }
+    elsif ( $script eq "read_2bit" )                { script_read_2bit(                 $in, $out, $options ) }
+    elsif ( $script eq "read_solexa" )              { script_read_solexa(               $in, $out, $options ) }
+    elsif ( $script eq "read_solid" )               { script_read_solid(                $in, $out, $options ) }
+    elsif ( $script eq "read_mysql" )               { script_read_mysql(                $in, $out, $options ) }
+    elsif ( $script eq "format_genome" )            { script_format_genome(             $in, $out, $options ) }
+    elsif ( $script eq "length_seq" )               { script_length_seq(                $in, $out, $options ) }
+    elsif ( $script eq "uppercase_seq" )            { script_uppercase_seq(             $in, $out, $options ) }
+    elsif ( $script eq "shuffle_seq" )              { script_shuffle_seq(               $in, $out, $options ) }
+    elsif ( $script eq "analyze_seq" )              { script_analyze_seq(               $in, $out, $options ) }
+    elsif ( $script eq "analyze_tags" )             { script_analyze_tags(              $in, $out, $options ) }
+    elsif ( $script eq "complexity_seq" )           { script_complexity_seq(            $in, $out, $options ) }
+    elsif ( $script eq "oligo_freq" )               { script_oligo_freq(                $in, $out, $options ) }
+    elsif ( $script eq "create_weight_matrix" )     { script_create_weight_matrix(      $in, $out, $options ) }
+    elsif ( $script eq "calc_bit_scores" )          { script_calc_bit_scores(           $in, $out, $options ) }
+    elsif ( $script eq "reverse_seq" )              { script_reverse_seq(               $in, $out, $options ) }
+    elsif ( $script eq "complement_seq" )           { script_complement_seq(            $in, $out, $options ) }
+    elsif ( $script eq "remove_indels" )            { script_remove_indels(             $in, $out, $options ) }
+    elsif ( $script eq "transliterate_seq" )        { script_transliterate_seq(         $in, $out, $options ) }
+    elsif ( $script eq "transliterate_vals" )       { script_transliterate_vals(        $in, $out, $options ) }
+    elsif ( $script eq "translate_seq" )            { script_translate_seq(             $in, $out, $options ) }
+    elsif ( $script eq "extract_seq" )              { script_extract_seq(               $in, $out, $options ) }
+    elsif ( $script eq "get_genome_seq" )           { script_get_genome_seq(            $in, $out, $options ) }
+    elsif ( $script eq "get_genome_align" )         { script_get_genome_align(          $in, $out, $options ) }
+    elsif ( $script eq "get_genome_phastcons" )     { script_get_genome_phastcons(      $in, $out, $options ) }
+    elsif ( $script eq "fold_seq" )                 { script_fold_seq(                  $in, $out, $options ) }
+    elsif ( $script eq "split_seq" )                { script_split_seq(                 $in, $out, $options ) }
+    elsif ( $script eq "split_bed" )                { script_split_bed(                 $in, $out, $options ) }
+    elsif ( $script eq "align_seq" )                { script_align_seq(                 $in, $out, $options ) }
+    elsif ( $script eq "tile_seq" )                 { script_tile_seq(                  $in, $out, $options ) }
+    elsif ( $script eq "invert_align" )             { script_invert_align(              $in, $out, $options ) }
+    elsif ( $script eq "patscan_seq" )              { script_patscan_seq(               $in, $out, $options ) }
+    elsif ( $script eq "create_blast_db" )          { script_create_blast_db(           $in, $out, $options ) }
+    elsif ( $script eq "blast_seq" )                { script_blast_seq(                 $in, $out, $options ) }
+    elsif ( $script eq "blat_seq" )                 { script_blat_seq(                  $in, $out, $options ) }
+    elsif ( $script eq "match_seq" )                { script_match_seq(                 $in, $out, $options ) }
+    elsif ( $script eq "create_vmatch_index" )      { script_create_vmatch_index(       $in, $out, $options ) }
+    elsif ( $script eq "vmatch_seq" )               { script_vmatch_seq(                $in, $out, $options ) }
+    elsif ( $script eq "write_fasta" )              { script_write_fasta(               $in, $out, $options ) }
+    elsif ( $script eq "write_align" )              { script_write_align(               $in, $out, $options ) }
+    elsif ( $script eq "write_blast" )              { script_write_blast(               $in, $out, $options ) }
+    elsif ( $script eq "write_tab" )                { script_write_tab(                 $in, $out, $options ) }
+    elsif ( $script eq "write_bed" )                { script_write_bed(                 $in, $out, $options ) }
+    elsif ( $script eq "write_psl" )                { script_write_psl(                 $in, $out, $options ) }
+    elsif ( $script eq "write_2bit" )               { script_write_2bit(                $in, $out, $options ) }
+    elsif ( $script eq "write_solid" )              { script_write_solid(               $in, $out, $options ) }
+    elsif ( $script eq "head_records" )             { script_head_records(              $in, $out, $options ) }
+    elsif ( $script eq "remove_keys" )              { script_remove_keys(               $in, $out, $options ) }
+    elsif ( $script eq "rename_keys" )              { script_rename_keys(               $in, $out, $options ) }
+    elsif ( $script eq "uniq_vals" )                { script_uniq_vals(                 $in, $out, $options ) }
+    elsif ( $script eq "merge_vals" )               { script_merge_vals(                $in, $out, $options ) }
+    elsif ( $script eq "grab" )                     { script_grab(                      $in, $out, $options ) }
+    elsif ( $script eq "compute" )                  { script_compute(                   $in, $out, $options ) }
+    elsif ( $script eq "flip_tab" )                 { script_flip_tab(                  $in, $out, $options ) }
+    elsif ( $script eq "add_ident" )                { script_add_ident(                 $in, $out, $options ) }
+    elsif ( $script eq "count_records" )            { script_count_records(             $in, $out, $options ) }
+    elsif ( $script eq "random_records" )           { script_random_records(            $in, $out, $options ) }
+    elsif ( $script eq "sort_records" )             { script_sort_records(              $in, $out, $options ) }
+    elsif ( $script eq "count_vals" )               { script_count_vals(                $in, $out, $options ) }
+    elsif ( $script eq "plot_histogram" )           { script_plot_histogram(            $in, $out, $options ) }
+    elsif ( $script eq "plot_lendist" )             { script_plot_lendist(              $in, $out, $options ) }
+    elsif ( $script eq "plot_chrdist" )             { script_plot_chrdist(              $in, $out, $options ) }
+    elsif ( $script eq "plot_karyogram" )           { script_plot_karyogram(            $in, $out, $options ) }
+    elsif ( $script eq "plot_matches" )             { script_plot_matches(              $in, $out, $options ) }
+    elsif ( $script eq "plot_seqlogo" )             { script_plot_seqlogo(              $in, $out, $options ) }
+    elsif ( $script eq "plot_phastcons_profiles" )  { script_plot_phastcons_profiles(   $in, $out, $options ) }
+    elsif ( $script eq "analyze_bed" )              { script_analyze_bed(               $in, $out, $options ) }
+    elsif ( $script eq "analyze_vals" )             { script_analyze_vals(              $in, $out, $options ) }
+    elsif ( $script eq "length_vals" )              { script_length_vals(               $in, $out, $options ) }
+    elsif ( $script eq "sum_vals" )                 { script_sum_vals(                  $in, $out, $options ) }
+    elsif ( $script eq "mean_vals" )                { script_mean_vals(                 $in, $out, $options ) }
+    elsif ( $script eq "median_vals" )              { script_median_vals(               $in, $out, $options ) }
+    elsif ( $script eq "max_vals" )                 { script_max_vals(                  $in, $out, $options ) }
+    elsif ( $script eq "min_vals" )                 { script_min_vals(                  $in, $out, $options ) }
+    elsif ( $script eq "upload_to_ucsc" )           { script_upload_to_ucsc(            $in, $out, $options ) }
 
     close $in if defined $in;
     close $out;
@@ -311,6 +312,13 @@ sub get_options
             num|n=s
         );
     }
+    elsif ( $script eq "read_fixedstep" )
+    {
+        @options = qw(
+            data_in|i=s
+            num|n=s
+        );
+    }
     elsif ( $script eq "read_blast_tab" )
     {
         @options = qw(
@@ -1346,6 +1354,59 @@ sub script_read_bed
 }
 
 
+sub script_read_fixedstep
+{
+    # Martin A. Hansen, Juli 2008.
+
+    # Read fixedStep wiggle format from stream or file.
+
+    my ( $in,        # handle to in stream
+         $out,       # handle to out stream
+         $options,   # options hash
+       ) = @_;
+
+    # Returns nothing.
+
+    my ( $file, $record, $entry, $head, $chr, $chr_beg, $step, $data_in, $num );
+
+    while ( $record = get_record( $in ) ) {
+        put_record( $record, $out );
+    }
+
+    $num = 1;
+
+    foreach $file ( @{ $options->{ "files" } } )
+    {
+        $data_in = Maasha::Common::read_open( $file );
+
+        while ( $entry = Maasha::UCSC::fixedstep_get_entry( $data_in ) )
+        {
+            $head = shift @{ $entry };
+
+            if ( $head =~ /^chrom=([^ ]+) start=(\d+) step=(\d+)$/ )
+            {
+                $record->{ "CHR" }     = $1;
+                $record->{ "CHR_BEG" } = $2;
+                $record->{ "STEP" }    = $3;
+                $record->{ "VALS" }    = join ",", @{ $entry };
+            }
+
+            put_record( $record, $out );
+
+            goto NUM if $options->{ "num" } and $num == $options->{ "num" };
+
+            $num++;
+        }
+
+        close $data_in;
+    }
+
+    NUM:
+
+    close $data_in if $data_in;
+}
+
+
 sub script_read_blast_tab
 {
     # Martin A. Hansen, September 2007.
@@ -1592,7 +1653,7 @@ sub script_read_phastcons
     {
         $data_in = Maasha::Common::read_open( $file );
 
-        while ( $entry = Maasha::UCSC::phastcons_get_entry( $data_in ) ) 
+        while ( $entry = Maasha::UCSC::fixedstep_get_entry( $data_in ) ) 
         {
             @records = Maasha::UCSC::phastcons_parse_entry( $entry, $options );
 

diff --git a/code_perl/Maasha/UCSC.pm b/code_perl/Maasha/UCSC.pm
index 8e02ddece4ee44c0f15c0fa8b3f82458923f06ef..4f932e3b5135de092033a663118ee75f4346bb4c 100644 (file)
--- a/code_perl/Maasha/UCSC.pm
+++ b/code_perl/Maasha/UCSC.pm
@@ -834,10 +834,10 @@ sub update_my_tracks
 }
 
 
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> PhastCons format <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> fixedStep format <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-sub phastcons_get_entry
+sub fixedstep_get_entry
 {
     # Martin A. Hansen, December 2007.
 
@@ -868,6 +868,9 @@ sub phastcons_get_entry
 }
 
 
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> PhastCons format <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+
+
 sub phastcons_parse_entry
 {
     # Martin A. Hansen, December 2007.
@@ -1023,7 +1026,7 @@ sub phastcons_index_create
 
     $pos = 0;
 
-    while ( $entry = Maasha::UCSC::phastcons_get_entry( $fh ) )
+    while ( $entry = Maasha::UCSC::fixedstep_get_entry( $fh ) )
     {
         $locator = shift @{ $entry };