feat(salmon): download and build for salmon v1.4.0

CHANGELOG.md

@@ -35,6 +35,7 @@ pdfmerger: v1.0
 picardtools: 2.23.4 -> 2.25.0
 preseq: 2.0.3 -> 3.1.2
 rseqc: 3.0.1 -> 4.0.0
+salmon: 0.12.0 -> 1.4.0
 samtools: 1.10-h9402c20_2 -> 1.12=h9aed4be_1
 smncopynumbercaller: 4b2c1ad -> v1.1.1
 star: 2.7.4a -> 2.7.8a

definitions/download_parameters.yaml

@@ -196,6 +196,13 @@ gencode_annotation:
   data_type: SCALAR
   default: 1
   type: recipe
+gencode_transcript:
+  analysis_mode: case
+  associated_recipe:
+    - mip
+  data_type: SCALAR
+  default: 1
+  type: recipe
 genomic_superdups:
   analysis_mode: case
   associated_recipe:
@@ -333,6 +340,7 @@ recipe_core_number:
     gatk_mitochondrial_ref: 1
     genbank_haplogroup: 1
     gencode_annotation: 1
+    gencode_transcript: 1
     genomic_superdups: 1
     giab: 1
     gnomad: 1
@@ -384,6 +392,7 @@ recipe_time:
     gatk_mitochondrial_ref: 1
     genbank_haplogroup: 1
     gencode_annotation: 1
+    gencode_transcript: 1
     genomic_superdups: 1
     giab: 1
     gnomad: 25

definitions/rd_rna_parameters.yaml

@@ -66,7 +66,6 @@ transcript_annotation:
     - gffcompare_ar
     - picardtools_collectrnaseqmetrics
     - rseqc
-    - salmon_quant
     - star_aln
     - star_fusion
     - stringtie_ar
@@ -86,6 +85,14 @@ transcript_annotation_file_endings:
   mandatory: no
   reference: reference_dir
   type: mip
+transcript_sequence:
+  associated_recipe:
+    - salmon_quant
+  data_type: SCALAR
+  exists_check: file
+  is_reference: 1
+  reference: reference_dir
+  type: path
 library_type:
   associated_recipe:
     - genebody_coverage
@@ -308,6 +315,12 @@ salmon_quant_reference_genome:
   mandatory: no
   reference: reference_dir
   type: path
+salmon_quant_gencode_reference:
+  associated_recipe:
+    - salmon_quant
+  data_type: SCALAR
+  default: 1
+  type: recipe_argument
 ## Blobfish
 blobfish:
   analysis_mode: case

documentation/Setup.md

@@ -66,7 +66,7 @@ You can speed up, for instance, the Readonly module by also installing the compa
 - [Rhocall] (version: 0.5.1)
 - [RSeQC] (version: 4.0.0)
 - [rtg-tools] (version: 3.10.1)
-- [Salmon] (version: 0.12.0)
+- [Salmon] (version: 1.4.0)
 - [Sambamba] (version: 0.6.8)
 - [Samtools] (version: 1.11)
 - [SMNCopyNumberCaller] (version: v1.1.1)

lib/MIP/Cli/Mip/Analyse/Rd_rna.pm

@@ -415,6 +415,14 @@ q{Default: grch37_dbsnp_-138-.vcf, grch37_1000g_indels_-phase1-.vcf, grch37_mill
         )
     );
 
+    option(
+        q{salmon_quant_gencode_reference} => (
+            documentation => q{Salmon uses a gencode reference},
+            is            => q{rw},
+            isa           => Bool,
+        )
+    );
+
     option(
         q{star_aln} => (
             cmd_tags      => [q{Analysis recipe switch}],
@@ -924,6 +932,15 @@ q{Regular expression file containing the regular expression to be used for each
         )
     );
 
+    option(
+        q{transcript_sequence} => (
+            cmd_tags      => [q{Transcript sequence file: Format: fasta}],
+            documentation => q{Transcript sequences for the rd_rna pipeline},
+            is            => q{rw},
+            isa           => Str,
+        )
+    );
+
     option(
         q{trim_galore_ar} => (
             cmd_tags      => [q{Analysis recipe switch}],

lib/MIP/Program/Gnu/Coreutils.pm

@@ -339,7 +339,8 @@ sub gnu_cut {
 ## Function : Perl wrapper for writing cut command to already open $filehandle or return commands array. Based on cut 8.4
 ## Returns  : @commands
 ## Arguments: $filehandle             => Filehandle to write to
-##          : $infile_path            => Infile paths {REF}
+##          : $delimiter              => Delimiter
+##          : $infile_path            => Infile path
 ##          : $list                   => List of specified fields
 ##          : $stderrfile_path        => Stderrfile path
 ##          : $stderrfile_path_append => Append to stderrinfo to file
@@ -348,6 +349,7 @@ sub gnu_cut {
     my ($arg_href) = @_;
 
     ## Flatten argument(s)
+    my $delimiter;
     my $filehandle;
     my $list;
     my $infile_path;
@@ -359,13 +361,16 @@ sub gnu_cut {
         filehandle => {
             store => \$filehandle,
         },
+        delimiter => {
+            store       => \$delimiter,
+            strict_type => 1,
+        },
         list => {
             store       => \$list,
             strict_type => 1,
         },
         infile_path => {
             defined     => 1,
-            required    => 1,
             store       => \$infile_path,
             strict_type => 1,
         },
@@ -388,13 +393,20 @@ sub gnu_cut {
     ## Stores commands depending on input parameters
     my @commands = qw{ cut };
 
+    if ($delimiter) {
+
+        push @commands, q{-d} . $SPACE . $delimiter;
+    }
+
     if ($list) {
 
         push @commands, q{-f} . $SPACE . $list;
     }
 
-    ## Infiles
-    push @commands, $infile_path;
+    if ($infile_path) {
+
+        push @commands, $infile_path;
+    }
 
     push @commands,
       unix_standard_streams(
@@ -505,8 +517,7 @@ sub gnu_echo {
     }
 
     ## Strings
-    push @commands,
-      $string_wrapper . join( $EMPTY_STR, @{$strings_ref} ) . $string_wrapper;
+    push @commands, $string_wrapper . join( $EMPTY_STR, @{$strings_ref} ) . $string_wrapper;
 
     if ($outfile_path) {
         push @commands, q{>} . $SPACE . $outfile_path;

lib/MIP/Program/Salmon.pm

@@ -32,26 +32,38 @@ Readonly my $BASE_COMMAND => q{salmon};
 
 sub salmon_index {
 
-## Function  : Perl wrapper for Salmon index, version 0.9.1.
+## Function  : Perl wrapper for Salmon index, version 1.4.0.
 ## Returns   : @commands
-## Arguments : $fasta_path             => Input reference fasta path, note salmon does not use the genome reference fasta, it uses a fasta file of transcripts
+## Arguments : $decoy_path             => Decoy sequence ids
+##           : $fasta_path             => Input reference fasta path, note salmon does not use the genome reference fasta, it uses a fasta file of transcripts
 ##           : $filehandle             => Filehandle to write to
+##           : $gencode                => Transcripts are in gencode format
 ##           : $outfile_path           => Outfile path
 ##           : $stderrfile_path        => Stderrfile path
 ##           : $stderrfile_path_append => Append stderr info to file path
 ##           : $stdoutfile_path        => Stdoutfile path
+##           : $temp_directory         => Temporary directory
+##           : $threads                => Threads used for indexing
 
     my ($arg_href) = @_;
 
     ## Flatten argument(s)
+    my $decoy_path;
     my $fasta_path;
     my $filehandle;
+    my $gencode;
     my $outfile_path;
     my $stderrfile_path;
     my $stderrfile_path_append;
     my $stdoutfile_path;
+    my $temp_directory;
+    my $threads;
 
     my $tmpl = {
+        decoy_path => {
+            store       => \$decoy_path,
+            strict_type => 1,
+        },
         fasta_path => {
             defined     => 1,
             required    => 1,
@@ -61,6 +73,11 @@ sub salmon_index {
         filehandle => {
             store => \$filehandle,
         },
+        gencode => {
+            allow       => [ undef, 0, 1 ],
+            store       => \$gencode,
+            strict_type => 1,
+        },
         outfile_path => {
             defined     => 1,
             required    => 1,
@@ -79,6 +96,15 @@ sub salmon_index {
             store       => \$stdoutfile_path,
             strict_type => 1,
         },
+        temp_directory => {
+            store       => \$temp_directory,
+            strict_type => 1,
+        },
+        threads => {
+            allow       => [ undef, qr/\A \d+ \z/xms ],
+            store       => \$threads,
+            strict_type => 1,
+        },
     };
 
     check( $tmpl, $arg_href, 1 ) or croak q{Could not parse arguments!};
@@ -90,6 +116,25 @@ sub salmon_index {
 
     push @commands, q{--index} . $SPACE . $outfile_path;
 
+    if ($decoy_path) {
+
+        push @commands, q{--decoy} . $SPACE . $decoy_path;
+    }
+
+    if ($gencode) {
+
+        push @commands, q{--gencode};
+    }
+
+    if ($temp_directory) {
+
+        push @commands, q{--tmpdir} . $SPACE . $temp_directory;
+    }
+
+    if ($threads) {
+        push @commands, q{--threads} . $SPACE . $threads,;
+    }
+
     push @commands,
       unix_standard_streams(
         {

lib/MIP/Recipes/Build/Salmon_quant_prerequisites.pm

@@ -16,7 +16,7 @@ use autodie qw{ :all };
 use Readonly;
 
 ## MIPs lib/
-use MIP::Constants qw{ $DOT $NEWLINE $UNDERSCORE };
+use MIP::Constants qw{ $NEWLINE $PIPE $SPACE $UNDERSCORE };
 
 BEGIN {
 
@@ -144,7 +144,9 @@ sub build_salmon_quant_prerequisites {
     use MIP::Program::Gnu::Coreutils qw{ gnu_mkdir };
     use MIP::Language::Shell qw{ check_exist_and_move_file };
     use MIP::Processmanagement::Processes qw{ submit_recipe };
-    use MIP::Program::Star_fusion qw{ star_fusion_gtf_file_to_feature_seqs };
+    use MIP::Program::Gnu::Coreutils qw{ gnu_cat gnu_cut };
+    use MIP::Program::Gnu::Software::Gnu_grep qw{ gnu_grep };
+    use MIP::Program::Gnu::Software::Gnu_sed qw{ gnu_sed };
     use MIP::Program::Salmon qw{ salmon_index };
     use MIP::Recipe qw{ parse_recipe_prerequisites };
     use MIP::Script::Setup_script qw{ setup_script };
@@ -206,24 +208,60 @@ sub build_salmon_quant_prerequisites {
     );
     say {$filehandle} $NEWLINE;
 
-    ## Build cDNA sequence file
-    star_fusion_gtf_file_to_feature_seqs(
+    ## Make decoy file
+    my $decoy_file_path = catfile( $salmon_quant_directory_tmp, q{decoys.txt} );
+    gnu_grep(
         {
-            filehandle         => $filehandle,
-            gtf_path           => $active_parameter_href->{transcript_annotation},
-            referencefile_path => $human_genome_reference,
-            seq_type           => q{cDNA},
-            stdoutfile_path    => catfile( $salmon_quant_directory_tmp, q{cDNA_seqs.fa} ),
+            filehandle  => $filehandle,
+            infile_path => $active_parameter_href->{transcript_sequence},
+            pattern     => q{'^>'},
+        }
+    );
+    print {$filehandle} $PIPE . $SPACE;
+
+    gnu_cut(
+        {
+            delimiter       => q{' '},
+            filehandle      => $filehandle,
+            list            => 1,
+            stdoutfile_path => $decoy_file_path,
+        }
+    );
+    say {$filehandle} $NEWLINE;
+
+    gnu_sed(
+        {
+            filehandle   => $filehandle,
+            infile_path  => $decoy_file_path,
+            inplace_edit => 1,
+            script       => q{'s/>//g'},
+        }
+    );
+    say {$filehandle} $NEWLINE;
+
+    ## Combine transcriptome and genome
+    gnu_cat(
+        {
+            filehandle       => $filehandle,
+            infile_paths_ref => [
+                $active_parameter_href->{transcript_sequence},
+                $active_parameter_href->{human_genome_reference}
+            ],
+            stdoutfile_path => catfile( $salmon_quant_directory_tmp, q{gentrome.fa} ),
         }
     );
     say {$filehandle} $NEWLINE;
 
     ## Build Salmon index file
     salmon_index(
         {
-            fasta_path   => catfile( $salmon_quant_directory_tmp, q{cDNA_seqs.fa} ),
-            filehandle   => $filehandle,
-            outfile_path => $salmon_quant_directory_tmp,
+            fasta_path     => catfile( $salmon_quant_directory_tmp, q{gentrome.fa} ),
+            threads        => $NUMBER_OF_CORES,
+            filehandle     => $filehandle,
+            temp_directory => $temp_directory,
+            gencode        => $active_parameter_href->{salmon_quant_gencode_reference},
+            decoy_path     => $decoy_file_path,
+            outfile_path   => $salmon_quant_directory_tmp,
         }
     );
     say {$filehandle} $NEWLINE;