|
|
@@ -60,123 +60,123 @@ my $global_options = checkParams(); |
|
|
# CODE HERE
|
|
|
######################################################################
|
|
|
|
|
|
-# # check that the file exists
|
|
|
-# checkFileExists($global_options->{'in'});
|
|
|
-
|
|
|
-# # run prokka to generate the ORFs and also prokka annotations
|
|
|
-# checkAndRunCommand("prokka", [{
|
|
|
- # "--locustag" => $global_options->{'locustag'},
|
|
|
- # "--outdir" => "prokka_annotation",
|
|
|
- # "--prefix" => $global_options->{'locustag'},
|
|
|
- # "--kingdom" => $global_options->{'kingdom'},
|
|
|
- # "--cpus" => $global_options->{'threads'},
|
|
|
- # "--keep_names",
|
|
|
- # $global_options->{'in'},
|
|
|
- # }], DIE_ON_FAILURE);
|
|
|
+# check that the file exists
|
|
|
+checkFileExists($global_options->{'in'});
|
|
|
+
|
|
|
+# run prokka to generate the ORFs and also prokka annotations
|
|
|
+checkAndRunCommand("prokka", [{
|
|
|
+ "--locustag" => $global_options->{'locustag'},
|
|
|
+ "--outdir" => "prokka_annotation",
|
|
|
+ "--prefix" => $global_options->{'locustag'},
|
|
|
+ "--kingdom" => $global_options->{'kingdom'},
|
|
|
+ "--cpus" => $global_options->{'threads'},
|
|
|
+ "--keep_names",
|
|
|
+ $global_options->{'in'},
|
|
|
+ }], DIE_ON_FAILURE);
|
|
|
|
|
|
# identify the ORF called amino acid fasta file for blast-ing
|
|
|
my $locus = $global_options->{'locustag'};
|
|
|
|
|
|
-# # blast against img
|
|
|
-# if (! -e "./$locus.faaVSimg.blastp")
|
|
|
-# {
|
|
|
-# print "BLASTing against IMG 4.0 database...............\n";
|
|
|
-# checkAndRunCommand("cat",
|
|
|
- # [[
|
|
|
- # "prokka_annotation/$locus.faa |",
|
|
|
- # "parallel",
|
|
|
- # "--block"=> "100k",
|
|
|
- # "--recstart",
|
|
|
- # "'>'",
|
|
|
- # "--pipe",
|
|
|
- # "blastp",
|
|
|
- # -db => "/srv/db/img/4.0/dereplicated/img_dereplicated_species.genes.faa",
|
|
|
- # -outfmt => 6,
|
|
|
- # -max_target_seqs => 1,
|
|
|
- # -evalue => $global_options->{'evalue'},
|
|
|
- # -query => "-",
|
|
|
- # "> $locus.faaVSimg.blastp",
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-# }
|
|
|
-
|
|
|
-# # reciprocal blast of img positive hits against genome ORF
|
|
|
-# if (! -e "./subsetimg.faaVS$locus.faa.blastp")
|
|
|
-# {
|
|
|
-# print "Reciprocal BLASTing positive IMG hits to $locus.faa ...............\n";
|
|
|
-# checkAndRunCommand("contig_extractor.pl",
|
|
|
- # [[
|
|
|
- # -i => "$locus.faaVSimg.blastp",
|
|
|
- # -d => "/srv/db/img/4.0/dereplicated/img_dereplicated_species.genes.faa",
|
|
|
- # -b => '',
|
|
|
- # -S => '',
|
|
|
- # -o => "subsetimg.faa",
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-
|
|
|
-# checkAndRunCommand("makeblastdb",
|
|
|
- # [[
|
|
|
- # -in => "prokka_annotation/$locus.faa",
|
|
|
- # -dbtype => "prot",
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-
|
|
|
-# checkAndRunCommand("blastp",
|
|
|
- # [[
|
|
|
- # -query => "subsetimg.faa",
|
|
|
- # -db => "prokka_annotation/$locus.faa",
|
|
|
- # -outfmt => 6,
|
|
|
- # -max_target_seqs => 1,
|
|
|
- # -evalue => $global_options->{'evalue'},
|
|
|
- # -num_threads => $global_options->{'threads'},
|
|
|
- # -out => "subsetimg.faaVS$locus.faa.blastp",
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-# }
|
|
|
-
|
|
|
-# # blast against uniref
|
|
|
-# if (! -e "./$locus.faaVSuniref90.blastp")
|
|
|
-# {
|
|
|
-# print "BLASTing against Uniref90 database................\n";
|
|
|
-# checkAndRunCommand("cat",[[
|
|
|
- # "prokka_annotation/$locus.faa |",
|
|
|
- # "parallel",
|
|
|
- # "--block"=> "100k",
|
|
|
- # "--recstart",
|
|
|
- # "'>'",
|
|
|
- # "--pipe",
|
|
|
- # "blastp",
|
|
|
- # -db => "/srv/db/uniprot/uniref-20140403/uniref90.fasta",
|
|
|
- # -outfmt => 6,
|
|
|
- # -max_target_seqs => 1,
|
|
|
- # -evalue => $global_options->{'evalue'},
|
|
|
- # -query => "-",
|
|
|
- # "> $locus.faaVSuniref90.blastp",
|
|
|
- # #-num_threads => $global_options->{'threads'},
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-# }
|
|
|
-
|
|
|
-# # reciprocal blast of Uniref positive hits against genome ORF
|
|
|
-# if (! -e "./subsetuniref.faaVS$locus.faa.blastp")
|
|
|
-# {
|
|
|
-# print "Reciprocal BLASTing positive Uniref hits to $locus.faa ...............\n";
|
|
|
-
|
|
|
-# checkAndRunCommand("contig_extractor.pl",
|
|
|
- # [[
|
|
|
- # -i => "$locus.faaVSuniref90.blastp",
|
|
|
- # -d => "/srv/db/uniprot/uniref-20140403/uniref90.fasta",
|
|
|
- # -b => '',
|
|
|
- # -S => '',
|
|
|
- # -o => "subsetuniref.faa",
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-
|
|
|
-# checkAndRunCommand("blastp",
|
|
|
- # [[
|
|
|
- # -query => "subsetuniref.faa",
|
|
|
- # -db => "prokka_annotation/$locus.faa",
|
|
|
- # -outfmt => 6,
|
|
|
- # -max_target_seqs => 1,
|
|
|
- # -evalue => $global_options->{'evalue'},
|
|
|
- # -num_threads => $global_options->{'threads'},
|
|
|
- # -out => "subsetuniref.faaVS$locus.faa.blastp",
|
|
|
- # ]], DIE_ON_FAILURE);
|
|
|
-# }
|
|
|
+# blast against img
|
|
|
+if (! -e "./$locus.faaVSimg.blastp")
|
|
|
+{
|
|
|
+print "BLASTing against IMG 4.0 database...............\n";
|
|
|
+checkAndRunCommand("cat",
|
|
|
+ [[
|
|
|
+ "prokka_annotation/$locus.faa |",
|
|
|
+ "parallel",
|
|
|
+ "--block"=> "100k",
|
|
|
+ "--recstart",
|
|
|
+ "'>'",
|
|
|
+ "--pipe",
|
|
|
+ "blastp",
|
|
|
+ -db => "/srv/db/img/4.0/dereplicated/img_dereplicated_species.genes.faa",
|
|
|
+ -outfmt => 6,
|
|
|
+ -max_target_seqs => 1,
|
|
|
+ -evalue => $global_options->{'evalue'},
|
|
|
+ -query => "-",
|
|
|
+ "> $locus.faaVSimg.blastp",
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+}
|
|
|
+
|
|
|
+# reciprocal blast of img positive hits against genome ORF
|
|
|
+if (! -e "./subsetimg.faaVS$locus.faa.blastp")
|
|
|
+{
|
|
|
+print "Reciprocal BLASTing positive IMG hits to $locus.faa ...............\n";
|
|
|
+checkAndRunCommand("contig_extractor.pl",
|
|
|
+ [[
|
|
|
+ -i => "$locus.faaVSimg.blastp",
|
|
|
+ -d => "/srv/db/img/4.0/dereplicated/img_dereplicated_species.genes.faa",
|
|
|
+ -b => '',
|
|
|
+ -S => '',
|
|
|
+ -o => "subsetimg.faa",
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+
|
|
|
+checkAndRunCommand("makeblastdb",
|
|
|
+ [[
|
|
|
+ -in => "prokka_annotation/$locus.faa",
|
|
|
+ -dbtype => "prot",
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+
|
|
|
+checkAndRunCommand("blastp",
|
|
|
+ [[
|
|
|
+ -query => "subsetimg.faa",
|
|
|
+ -db => "prokka_annotation/$locus.faa",
|
|
|
+ -outfmt => 6,
|
|
|
+ -max_target_seqs => 1,
|
|
|
+ -evalue => $global_options->{'evalue'},
|
|
|
+ -num_threads => $global_options->{'threads'},
|
|
|
+ -out => "subsetimg.faaVS$locus.faa.blastp",
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+}
|
|
|
+
|
|
|
+# blast against uniref
|
|
|
+if (! -e "./$locus.faaVSuniref90.blastp")
|
|
|
+{
|
|
|
+print "BLASTing against Uniref90 database................\n";
|
|
|
+checkAndRunCommand("cat",[[
|
|
|
+ "prokka_annotation/$locus.faa |",
|
|
|
+ "parallel",
|
|
|
+ "--block"=> "100k",
|
|
|
+ "--recstart",
|
|
|
+ "'>'",
|
|
|
+ "--pipe",
|
|
|
+ "blastp",
|
|
|
+ -db => "/srv/db/uniprot/uniref-20140403/uniref90.fasta",
|
|
|
+ -outfmt => 6,
|
|
|
+ -max_target_seqs => 1,
|
|
|
+ -evalue => $global_options->{'evalue'},
|
|
|
+ -query => "-",
|
|
|
+ "> $locus.faaVSuniref90.blastp",
|
|
|
+ #-num_threads => $global_options->{'threads'},
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+}
|
|
|
+
|
|
|
+# reciprocal blast of Uniref positive hits against genome ORF
|
|
|
+if (! -e "./subsetuniref.faaVS$locus.faa.blastp")
|
|
|
+{
|
|
|
+print "Reciprocal BLASTing positive Uniref hits to $locus.faa ...............\n";
|
|
|
+
|
|
|
+checkAndRunCommand("contig_extractor.pl",
|
|
|
+ [[
|
|
|
+ -i => "$locus.faaVSuniref90.blastp",
|
|
|
+ -d => "/srv/db/uniprot/uniref-20140403/uniref90.fasta",
|
|
|
+ -b => '',
|
|
|
+ -S => '',
|
|
|
+ -o => "subsetuniref.faa",
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+
|
|
|
+checkAndRunCommand("blastp",
|
|
|
+ [[
|
|
|
+ -query => "subsetuniref.faa",
|
|
|
+ -db => "prokka_annotation/$locus.faa",
|
|
|
+ -outfmt => 6,
|
|
|
+ -max_target_seqs => 1,
|
|
|
+ -evalue => $global_options->{'evalue'},
|
|
|
+ -num_threads => $global_options->{'threads'},
|
|
|
+ -out => "subsetuniref.faaVS$locus.faa.blastp",
|
|
|
+ ]], DIE_ON_FAILURE);
|
|
|
+}
|
|
|
|
|
|
# hashes for img
|
|
|
my %hash = ();
|
|
|
|
0 comments on commit
39bcdac