Update Kraken QC

README.md

@@ -42,7 +42,6 @@ ls Roary-*
 ```
 
 ###Add to your Environment
-
 Add the following lines to your $HOME/.bashrc file, or to /etc/profile.d/roary.sh to make it available to all users:
 
 ```
@@ -55,3 +54,15 @@ export PERL5LIB=$PERL5LIB:$HOME/Roary-x.x.x/lib
 ```
 cpanm Array::Utils BioPerl Exception::Class File::Find::Rule File::Grep File::Slurp::Tiny Graph Moose Moose::Role Text::CSV Log::Log4perl File::Which
 ```
+
+#When things go wrong
+###cdhit seg faults
+Old versions of cdhit have a bug, so you need to use at least version 4.6.1.  The cdhit packages for Ubuntu 12.04 seem to be effected, so [installing from the source](http://cd-hit.org/) is the only option. 
+
+###I installed the homebrew Kraken package and now theres an error when I run the tests or QC
+Theres a bug and you'll need to [install it from source](https://ccb.jhu.edu/software/kraken/) on older versions of OSX (like Mountain Lion).  
+
+###Why dont you bundle a Kraken database for the QC?
+Its massive (2.7GB) and changes as RefSeq is updated.  The [authors](https://ccb.jhu.edu/software/kraken/) have prebuilt databases and details about how to make your own.
+
+

dist.ini

@@ -1,5 +1,5 @@
 name    = Bio-Roary
-version = 2.2.1
+version = 2.2.2
 author  = Andrew J. Page <[email protected]>
 license = GPL_3
 copyright_holder = Wellcome Trust Sanger Institute
@@ -32,5 +32,10 @@ requires = parallel
 [Encoding]
 filename = t/data/expected_set_difference_common_set_plot.png
 filename = t/data/expected_set_difference_unique_set_two_plot.png
+filename = t/data/kraken_test/database.idx
+filename = t/data/kraken_test/database.jdb
+filename = t/data/kraken_test/database.kdb
+filename = t/data/kraken_test/taxonomy/names.dmp
+filename = t/data/kraken_test/taxonomy/nodes.dmp
 
 encoding = bytes

lib/Bio/Roary/CommandLine/CreatePanGenome.pm

@@ -65,8 +65,11 @@ sub usage_text {
     # Include full annotation and inference in group statistics
     create_pan_genome --verbose_stats *.gff
 
-    # Increase the groups/clusters limit (default 50,000). Please check the QC results before running this
+    # Increase the groups/clusters limit (default 50,000). Please check the QC results before running this!
     create_pan_genome --group_limit 60000  *.gff
+
+    # Use a different Kraken database
+    roary -k /path/to/kraken_database/  *.gff
 
     # This help message
     create_pan_genome -h

lib/Bio/Roary/CommandLine/Roary.pm

@@ -40,6 +40,7 @@ has 'translation_table'       => ( is => 'rw', isa => 'Int', default => 11 );
 has 'group_limit'             => ( is => 'rw', isa => 'Num', default => 50000 );
 has 'core_definition'         => ( is => 'rw', isa => 'Num', default => 1 );
 has 'verbose'                 => ( is => 'rw', isa => 'Bool', default => 0 );
+has 'kraken_db'               => ( is => 'rw', isa => 'Str',  default => '/lustre/scratch108/pathogen/pathpipe/kraken/minikraken_20140330/' );
 
 has 'run_qc' => ( is => 'rw', isa => 'Bool', default => 0 );
 
@@ -51,7 +52,7 @@ sub BUILD {
         $max_threads,           $dont_delete_files, $dont_split_groups,       $perc_identity, $output_filename,
         $job_runner,            $makeblastdb_exec,  $mcxdeblast_exec,         $mcl_exec,      $blastp_exec,
         $apply_unknowns_filter, $cpus,              $output_multifasta_files, $verbose_stats, $translation_table,
-        $run_qc,                $core_definition,   $help
+        $run_qc,                $core_definition,   $help, $kraken_db,
     );
 
     GetOptionsFromArray(
@@ -76,6 +77,7 @@ sub BUILD {
         'dont_run_qc'               => \$dont_run_qc,
         'cd|core_definition=i'      => \$core_definition,
         'v|verbose'                 => \$verbose,
+        'k|kraken_db=s'             => \$kraken_db,
         'h|help'                    => \$help,
     );
 
@@ -119,6 +121,7 @@ sub BUILD {
     $self->verbose_stats($verbose_stats)         if ( defined $verbose_stats );
     $self->translation_table($translation_table) if ( defined($translation_table) );
     $self->group_limit($group_limit)             if ( defined($group_limit) );
+    $self->kraken_db($kraken_db)                 if ( defined($kraken_db) );
 
 
     if ( defined($run_qc) ) {
@@ -167,7 +170,8 @@ sub run {
             input_files => $self->fasta_files,
             job_runner  => $self->job_runner,
             cpus        => $self->cpus,
-            verbose     => $self->verbose
+            verbose     => $self->verbose,
+            kraken_db   => $self->kraken_db
         );
         $qc_input_files->report;
     }
@@ -238,7 +242,7 @@ sub usage_text {
 
     # Generate QC report detailing top genus and species for each assembly
 	# Requires Kraken to be installed
-    roary -qc *.gff
+    roary -k /path/to/kraken_database/ -qc *.gff
 
     # This help message
     roary -h

lib/Bio/Roary/JobRunner/Local.pm

@@ -15,13 +15,17 @@ package Bio::Roary::JobRunner::Local;
 =cut
 
 use Moose;
+use Log::Log4perl qw(:easy);
 
 has 'commands_to_run' => ( is => 'ro', isa => 'ArrayRef', required => 1 );
+has 'logger'                  => ( is => 'ro', lazy => 1, builder => '_build_logger');
+has 'verbose'                 => ( is => 'rw', isa => 'Bool', default => 0 );
 
 sub run {
     my ($self) = @_;
 
     for my $command_to_run ( @{ $self->commands_to_run } ) {  
+        $self->logger->info($command_to_run);
         system($command_to_run );
     }
     1;
@@ -36,9 +40,23 @@ sub _construct_dependancy_params
 
 sub submit_dependancy_job {
     my ( $self,$command_to_run) = @_;
+    $self->logger->info($command_to_run);
     system($command_to_run );
 }
 
+sub _build_logger
+{
+    my ($self) = @_;
+    my $level = $ERROR;
+    if($self->verbose)
+    {
+       $level = $DEBUG;
+    }
+    Log::Log4perl->easy_init($level);
+    my $logger = get_logger();
+    return $logger;
+}
+
 no Moose;
 __PACKAGE__->meta->make_immutable;
 

lib/Bio/Roary/JobRunner/Parallel.pm

@@ -18,13 +18,20 @@ package Bio::Roary::JobRunner::Parallel;
 use Moose;
 use File::Slurp::Tiny qw(read_file write_file);
 use File::Temp qw/ tempfile /;
+use Log::Log4perl qw(:easy);
 
 has 'commands_to_run' => ( is => 'ro', isa => 'ArrayRef', required => 1 );
 has 'cpus'            => ( is => 'ro', isa => 'Int',      default => 1 );
+has 'logger'                  => ( is => 'ro', lazy => 1, builder => '_build_logger');
+has 'verbose'                 => ( is => 'rw', isa => 'Bool', default => 0 );
 
 sub run {
     my ($self) = @_;
 
+    for my $command_to_run(@{ $self->commands_to_run })
+    {
+       $self->logger->info($command_to_run);
+    }
 	open(my $fh,"|-","parallel -j ".$self->cpus) || die "GNU Parallel failed";
 	print $fh join("\n", @{ $self->commands_to_run });
 	close $fh;
@@ -39,9 +46,23 @@ sub _construct_dependancy_params
 
 sub submit_dependancy_job {
     my ( $self,$command_to_run) = @_;
+    $self->logger->info($command_to_run);
     system($command_to_run );
 }
 
+sub _build_logger
+{
+    my ($self) = @_;
+    my $level = $ERROR;
+    if($self->verbose)
+    {
+       $level = $DEBUG;
+    }
+    Log::Log4perl->easy_init($level);
+    my $logger = get_logger();
+    return $logger;
+}
+
 no Moose;
 __PACKAGE__->meta->make_immutable;
 

lib/Bio/Roary/JobRunner/Role.pm

@@ -19,11 +19,17 @@ has '_queue'                  => ( is => 'rw', isa => 'Str',  default => 'normal
 has 'dont_wait'               => ( is => 'rw', isa => 'Bool', default => 0 );
 has 'cpus'                    => ( is => 'ro', isa => 'Int',      default => 1 );
 has 'logger'                  => ( is => 'ro', lazy => 1, builder => '_build_logger');
+has 'verbose'                 => ( is => 'rw', isa => 'Bool', default => 0 );
 
 sub _build_logger
 {
     my ($self) = @_;
-    Log::Log4perl->easy_init($ERROR);
+    my $level = $ERROR;
+    if($self->verbose)
+    {
+       $level = $DEBUG;
+    }
+    Log::Log4perl->easy_init($level);
     my $logger = get_logger();
     return $logger;
 }