refactoring of assemble_pairs

[biopieces.git] / bp_bin / format_genome

diff --git a/bp_bin/format_genome b/bp_bin/format_genome
index af7de5c4ca25f3e18f5a64eeb85baff16a554055..9b3e4fe81790b7ea159efaad477bb5d87922dbad 100755 (executable)
--- a/bp_bin/format_genome
+++ b/bp_bin/format_genome
@@ -31,6 +31,7 @@ use strict;
 use Maasha::Fasta;
 use Maasha::Biopieces;
 use Maasha::Bowtie;
+use Maasha::BWA;
 use Maasha::NCBI;
 use Maasha::Match;
 use Maasha::UCSC;
@@ -44,7 +45,7 @@ my ( $default, $formats, $options, $in, $out, $record, $data_out, $entry,
 
 $tmp_dir = Maasha::Biopieces::get_tmpdir();
 $default = $ENV{ 'BP_DATA' };
-$formats = 'fasta,blast,vmatch,bowtie,phastcons';
+$formats = 'fasta,blast,vmatch,bowtie,bwa,phastcons';
 
 $options = Maasha::Biopieces::parse_options(
     [
@@ -64,7 +65,7 @@ $genome = $options->{ 'genome' };
 Maasha::Filesys::dir_create_if_not_exists( "$dir/genomes" );
 Maasha::Filesys::dir_create_if_not_exists( "$dir/genomes/$genome" );
 
-if ( grep { $_ =~ /fasta|blast|vmatch|bowtie/i } @{ $options->{ "formats" } } )
+if ( grep { $_ =~ /fasta|blast|vmatch|bowtie|bwa/i } @{ $options->{ "formats" } } )
 {
     if ( -f "$dir/genomes/$genome/fasta/$genome.fna" )
     {
@@ -94,7 +95,7 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
     {
         Maasha::Fasta::put_entry( $entry, $fh_out );
     }
-    elsif ( $fh_out and $record->{ "CHR" } and $record->{ "CHR_BEG" } and $record->{ "STEP" } and $record->{ "VALS" } )
+    elsif ( $fh_out and $record->{ "CHR" } and $record->{ "CHR_BEG" } and $record->{ "STEP" } and $record->{ "VALS" } )  # TODO: clean this!
     {
         print $fh_out "fixedStep chrom=$record->{ 'CHR' } start=$record->{ 'CHR_BEG' } step=$record->{ 'STEP' }\n";
 
@@ -108,22 +109,23 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
     Maasha::Biopieces::put_record( $record, $out ) if not $options->{ "no_stream" };
 }
 
+close $fh_out if $fh_out;
+
 foreach $format ( @{ $options->{ 'formats' } } )
 {
     print STDERR qq(Creating format: $format for $genome ... ) if $options->{ 'verbose' };
 
-    if    ( $format =~ /^fasta$/i )     { Maasha::Fasta::fasta_index( "$fasta_dir/$genome.fna", "$dir/genomes/$genome/fasta/$genome.index" ) }
+    if    ( $format =~ /^fasta$/i )     { Maasha::Fasta::fasta_index( "$fasta_dir/$genome.fna", $fasta_dir, "$genome.index" ) }
     elsif ( $format =~ /^blast$/i )     { Maasha::NCBI::blast_index( "$genome.fna", $fasta_dir, "$dir/genomes/$genome/blast", "dna", $genome ) }
-    elsif ( $format =~ /^blat$/i )      { warn "BLAT FORMAT NOT IMPLEMENTED" }
+    elsif ( $format =~ /^blat$/i )      { warn "BLAT FORMAT NOT IMPLEMENTED\n" }
     elsif ( $format =~ /^vmatch$/i )    { Maasha::Match::vmatch_index( "$genome.fna", $fasta_dir, "$dir/genomes/$genome/vmatch", $tmp_dir ) }
     elsif ( $format =~ /^bowtie$/i )    { Maasha::Bowtie::bowtie_index( "$fasta_dir/$genome.fna", "$dir/genomes/$genome/bowtie", $genome, $options->{ 'verbose' } ) }
+    elsif ( $format =~ /^bwa$/i )       { Maasha::BWA::bwa_index( "$fasta_dir/$genome.fna", "$dir/genomes/$genome/bwa", $genome, $options->{ 'verbose' } ) }
     elsif ( $format =~ /^phastcons$/i ) { Maasha::UCSC::phastcons_index( "$genome.pp", $phastcons_dir ) }
 
     print STDERR qq(done.\n) if $options->{ 'verbose' };
 }
 
-close $fh_out if $fh_out;
-
 Maasha::Biopieces::close_stream( $in );
 Maasha::Biopieces::close_stream( $out );