From 8c9c4078cbd2b21ca040350fd227640af94e618b Mon Sep 17 00:00:00 2001
From: Emmanuel Mongin <mongin@sanger.ac.uk>
Date: Thu, 12 Jun 2003 09:26:22 +0000
Subject: [PATCH] dros options
---
misc-scripts/protein_match/get_Xmapping.pl | 6 +++--
misc-scripts/protein_match/mapping_conf.pl | 26 +++++++++----------
.../protein_match/prepare_proteome.pl | 4 ++-
misc-scripts/protein_match/scanwisepep.pl | 2 +-
4 files changed, 21 insertions(+), 17 deletions(-)
diff --git a/misc-scripts/protein_match/get_Xmapping.pl b/misc-scripts/protein_match/get_Xmapping.pl
index e06ff06dfd..09b29f67a3 100644
--- a/misc-scripts/protein_match/get_Xmapping.pl
+++ b/misc-scripts/protein_match/get_Xmapping.pl
@@ -19,7 +19,7 @@ my $organism = $conf{'organism'};
my $sptr_swiss = $conf{'sptr_swiss'};
my $out = $conf{'x_map_out'};
-#Get specific options for human/mouse/rat
+#Get specific options for human/mouse/rat/drosophila
my $refseq_gnp = $conf{'refseq_gnp'};
#Get specific options for human
@@ -42,6 +42,8 @@ my $eleg_nom = $conf{'eleg_nom'};
my $zeb_gene = $conf{'zeb_gene'};
my $zeb_dblink = $conf{'zeb_dblink'};
+#Get specific options for drosophila
+my $ext_annot = $conf{'dros_ext_annot'};
my $briggsae_peptides = $conf{'briggsae_hybrid'};
@@ -63,7 +65,7 @@ my ($swiss, $ac, $id) = &parse_sp_file($sptr_swiss);
&process_parsed_sp($swiss, $ac, $id, \*OUT);
-if (($organism eq "human") || ($organism eq "mouse") || ($organism eq "rat") || ($organism eq "zebrafish")) {
+if (($organism eq "human") || ($organism eq "mouse") || ($organism eq "rat") || ($organism eq "zebrafish") || ($organism eq "drosophila")) {
#Read the refseq file in gnp format
print STDERR "Reading REFSEQ File\n";
diff --git a/misc-scripts/protein_match/mapping_conf.pl b/misc-scripts/protein_match/mapping_conf.pl
index bcbad2a83b..41f00ab97f 100644
--- a/misc-scripts/protein_match/mapping_conf.pl
+++ b/misc-scripts/protein_match/mapping_conf.pl
@@ -28,10 +28,10 @@ package main;
#The mapping to known genes is assymetrical. This is due to the fact that's our gene prediction is quite fragmented compared to the manually curated genes
#'query_idt' => 40,
- 'query_idt' => ,
+ 'query_idt' => 50,
#'target_idt => 10,
- 'target_idt' => ,
+ 'target_idt' => 50,
@@ -47,7 +47,7 @@ package main;
#Location of the query peptide file (eg: Ensembl predicted protein)
#'query' => '/work1/mongin/mapping/primary/ensembl110.pep',
- 'query' => '',
+ 'query' => '/acari/work4/mongin/dros3_mapping/Primary/drosophila-release3-peptides.fasta',
#Location of the sptr file, this file will be used as an
#input to grep the specific sp entries to the organism
@@ -58,32 +58,32 @@ package main;
#Location of the sptr file in fasta format containing the entries specific to the organism
#'sptr_fa' => '/work1/mongin/mapping/primary/HS.f',
- 'sptr_fa' => '',
+ 'sptr_fa' => '/acari/work4/mongin/dros3_mapping/Primary/7227.FASTAC',
#Location of the sptr file in Swiss-Prot format containing the entries specific to the organism
#'sptr_swiss' => '/ecs2/work1/lec/briggsae_peptides/briggsae.test',
- 'sptr_swiss' => '',
+ 'sptr_swiss' => '/acari/work4/mongin/dros3_mapping/Primary/7227.SPC',
#Location of the file containing all refseq and all SP in fasta format (This file will be produced by running prepare_proteome.pl)
- 'pmatch_input_fa' => '',
+ 'pmatch_input_fa' => '/acari/work4/mongin/dros3_mapping/Primary/total.fa',
#Output file containing the mapping of SP and refseq sequences to external databases
- 'x_map_out' => '',
+ 'x_map_out' => '/acari/work4/mongin/dros3_mapping/Output/xmap.out',
#Output file from pmatch.pl and input file for maps2db.pl
#'pmatch_out' => '/work1/mongin/mapping/outputs/pmatch_human1.txt',
- 'pmatch_out' => '',
+ 'pmatch_out' => '/acari/work4/mongin/dros3_mapping/Output/pmatch.out',
#Location of the Refseq (proteins) file in fasta format
#'refseq_fa' => '/work1/mongin/mapping/primary/refseq.fa',
- 'refseq_fa' => '',
+ 'refseq_fa' => '/acari/work4/mongin/dros3_mapping/Primary/fly.faa',
#Location of the Refseq (proteins) file in Genbank format
#'refseq_gnp' => '/work1/mongin/mouse/mapping/primary/mouse.gnp',
- 'refseq_gnp' => '',
+ 'refseq_gnp' => '/acari/work4/mongin/dros3_mapping/Primary/fly.gnp',
############################################
#Organism specific files for the X_mapping #
@@ -147,7 +147,7 @@ package main;
#drosophila#
############
- 'dros_ext_annot' => '',
+ 'dros_ext_annot' => '/acari/work4/mongin/dros3_mapping/Primary/xrefs.txt',
##########
@@ -184,7 +184,7 @@ package main;
#Location for pmatch binaries
#'pmatch' => '/nfs/disk65/ms2/bin/pmatch'
- 'pmatch' => '/nfs/disk65/ms2/bin/pmatch',
+ 'pmatch' => '/nfs/disk5/ms2/bin/pmatch',
@@ -195,7 +195,7 @@ package main;
#Name of the organism studied. Current keywords used(or planned to be used): human, drosophila, mouse, elegans, anopheles, zebrafish
#You can adapt the other scripts given the organisms (eg: do some specific x_mapping for a given organism)
#'organism' => 'human'
- 'organism' => '',
+ 'organism' => 'drosophila',
#OX (Organism taxonomy cross-reference) number
diff --git a/misc-scripts/protein_match/prepare_proteome.pl b/misc-scripts/protein_match/prepare_proteome.pl
index dc167c7886..afa8997a8c 100755
--- a/misc-scripts/protein_match/prepare_proteome.pl
+++ b/misc-scripts/protein_match/prepare_proteome.pl
@@ -59,7 +59,9 @@ my $organism = $conf{'organism'};
my $refseq_pred = $conf{'refseq_pred_fa'};
my $sub_genes = $conf{'submitted_genes'};
-if (($organism eq "human") || ($organism eq "mouse") || ($organism eq "rat")) {
+print STDERR "Prepare proteome using following files: SPTR ($sptr)\nREFSEQ ($refseq)\nPROTFILE ($protfile)\n";
+
+if (($organism eq "human") || ($organism eq "mouse") || ($organism eq "rat") || ($organism eq "drosophila")) {
&parse_refseq;
}
diff --git a/misc-scripts/protein_match/scanwisepep.pl b/misc-scripts/protein_match/scanwisepep.pl
index dcef766db6..548fc5593f 100644
--- a/misc-scripts/protein_match/scanwisepep.pl
+++ b/misc-scripts/protein_match/scanwisepep.pl
@@ -37,7 +37,7 @@ my %query2length;
#################################
# run scanwisepep
- my $scanwise = "/nfs/acari/birney/prog/wise2/src/models/scanwisep -seqdb $target $query -seqloadtile 5 -hspthread -hspthreadno 4 -hsp2hit_best -hsp2hit_best_perc 10 -hitoutput tab >> /tmp/$$.pmatch";
+ my $scanwise = "/usr/local/ensembl/bin/scanwisep-2.2.3-rc6 -seqdb $target $query -seqloadtile 5 -hspthread -hspthreadno 4 -hsp2hit_best -hsp2hit_best_perc 10 -hitoutput tab >> /tmp/$$.pmatch";
print STDERR "Running Scanwise: $scanwise\n";
--
GitLab