From 80b4e6bee8473463b0fcea83041b506611342b9f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Andreas=20Kusalananda=20K=C3=A4h=C3=A4ri?=
<ak4@sanger.ac.uk>
Date: Tue, 16 Nov 2010 14:10:13 +0000
Subject: [PATCH] Remove files no longer used.
---
.../unmapped_reason/unmapped_reason.txt | 51 -------
.../update_unmapped_reasons.pl | 131 ------------------
2 files changed, 182 deletions(-)
delete mode 100644 misc-scripts/unmapped_reason/unmapped_reason.txt
delete mode 100644 misc-scripts/unmapped_reason/update_unmapped_reasons.pl
diff --git a/misc-scripts/unmapped_reason/unmapped_reason.txt b/misc-scripts/unmapped_reason/unmapped_reason.txt
deleted file mode 100644
index e1f4ef5a2f..0000000000
--- a/misc-scripts/unmapped_reason/unmapped_reason.txt
+++ /dev/null
@@ -1,51 +0,0 @@
-1 Marker matches multiple times Marker aligns to the genome > 3 times
-2 Marker does not align Unable to align to the genome
-3 Failed to find Stable ID Stable ID that this xref was linked to no longer exists
-4 No mapping done No mapping done for this type of xref
-5 Failed to match Unable to match to any ensembl entity at all
-6 Failed to match at thresholds Unable to match at the thresholds of 90% for the query or 90% for the target
-7 No Master The dependent xref was not matched due to there being no master xref
-8 Master failed The dependent xref was not matched due to the master xref not being mapped
-66 Did not meet threshold Match score for transcript lower than threshold (0.75)
-65 Was not best match Did not top best transcript match score (1.00)
-64 Was not best match Did not top best transcript match score (0.99)
-63 Was not best match Did not top best transcript match score (0.98)
-62 Was not best match Did not top best transcript match score (0.97)
-61 Was not best match Did not top best transcript match score (0.96)
-60 Was not best match Did not top best transcript match score (0.95)
-59 Was not best match Did not top best transcript match score (0.94)
-58 Was not best match Did not top best transcript match score (0.93)
-57 Was not best match Did not top best transcript match score (0.92)
-56 Was not best match Did not top best transcript match score (0.91)
-55 Was not best match Did not top best transcript match score (0.90)
-54 Was not best match Did not top best transcript match score (0.89)
-53 Was not best match Did not top best transcript match score (0.88)
-52 Was not best match Did not top best transcript match score (0.87)
-51 Was not best match Did not top best transcript match score (0.86)
-50 Was not best match Did not top best transcript match score (0.85)
-49 Was not best match Did not top best transcript match score (0.84)
-48 Was not best match Did not top best transcript match score (0.83)
-47 Was not best match Did not top best transcript match score (0.82)
-46 Was not best match Did not top best transcript match score (0.81)
-45 Was not best match Did not top best transcript match score (0.80)
-44 Was not best match Did not top best transcript match score (0.79)
-43 Was not best match Did not top best transcript match score (0.78)
-42 Was not best match Did not top best transcript match score (0.77)
-41 Was not best match Did not top best transcript match score (0.76)
-40 Was not best match Did not top best transcript match score (0.75)
-67 No overlap No coordinate overlap with any Ensembl transcript
-96 Failed to match at thresholds Unable to match at the thresholds of 100% for the query or 100% for the target
-125 Failed to match at thresholds Unable to match at the thresholds of 55% for the query or 55% for the target
-126 >10% N-strings More than 10% of the sequence consists of strings of Ns. Sequences are not rejected for this reason but this may explain a low coverage hit
-127 All long introns Every intron in these hits is of length 250000-400000bp, we require at least one intron to be shorter than 250000bp
-128 GSS sequence This cDNA has been excluded from the analysis because it is in the GSS (Genome Survey Sequence) division of GenBank
-129 Low coverage Coverage of the best alignment is less than 90% - see query_score for coverage
-130 Low coverage with long intron Hits containing introns longer than 250000bp are rejected if coverage is less than 98% - see query_score for coverage
-131 Low percent_id with long intron Hits containing introns longer than 250000bp are rejected if percentage identity is less than 98% - see query_score for percent_id
-132 Low percent_id Percentage identity of the best alignment is less than 97% - see query_score for percent_id
-133 No output from Exonerate Exonerate returned no hits using standard parameters plus options --maxintron 400000 and --softmasktarget FALSE
-134 Parent xref failed to match Unable to match as parent xref was not mapped
-135 Processed pseudogene Rejected as a processed pseudogene because there are multiple-exon hits with the same coverage which have been rejected for other reasons
-136 See kill-list database This sequence has been excluded from the analysis - see the kill-list database for further details
-137 Failed to match at thresholds Unable to match at the thresholds of 99% for the query or 99% for the target
-138 Marker matches multiple times Marker aligns to the genome > 5 times
diff --git a/misc-scripts/unmapped_reason/update_unmapped_reasons.pl b/misc-scripts/unmapped_reason/update_unmapped_reasons.pl
deleted file mode 100644
index 272ea96bf5..0000000000
--- a/misc-scripts/unmapped_reason/update_unmapped_reasons.pl
+++ /dev/null
@@ -1,131 +0,0 @@
-#
-# updates the unmapped_reason tables on all of the core databases on a given host
-#
-
-
-use strict;
-
-use Getopt::Long;
-use DBI;
-use IO::File;
-
-my ( $host, $user, $pass, $port,@dbnames, $file, $release_num);
-
-GetOptions( "dbhost|host=s", \$host,
- "dbuser|user=s", \$user,
- "dbpass|pass=s", \$pass,
- "dbport|port=i", \$port,
- "file=s", \$file,
- "dbnames=s@", \@dbnames, # either provide -dbnames or -release
- "release_num=i", \$release_num
- );
-
-#both host and file are required
-usage() if(!$host || !$file);
-#release num XOR dbname are required
-if(($release_num && @dbnames) || (!$release_num && !@dbnames)) {
- print "\nYou can't use -dbnames <> and -release_num <> options at the same time\n" ;
- sleep(3) ;
- usage() ;
-}
-
-$port ||= 3306;
-
-my $dsn = "DBI:mysql:host=$host;port=$port";
-
-my $db = DBI->connect( $dsn, $user, $pass, {RaiseError => 1} );
-
-if($release_num) {
- @dbnames = map {$_->[0] } @{ $db->selectall_arrayref( "show databases" ) };
-
- #
- # filter out all non-core databases
- #
- @dbnames = grep {/^[a-zA-Z]+\_[a-zA-Z]+\_(core|est|estgene|vega)\_${release_num}\_\d+[A-Za-z]?$/} @dbnames;
-}
-
-
-#
-# make sure the user wishes to continue
-#
-print STDERR "The following databases will be unmapped_reason updated:\n ";
-print join("\n ", @dbnames);
-print "\ncontinue with update (yes/no)> ";
-
-my $input = lc(<STDIN>);
-chomp($input);
-if($input ne 'yes') {
- print "unmapped_reason conversion aborted\n";
- exit();
-}
-
-#
-# read all of the new external_db entries from the file
-#
-my $fh = IO::File->new();
-$fh->open($file) or die("could not open input file $file");
-my @rows;
-my $row;
-while($row = <$fh>) {
- chomp($row);
- my @a = split(/\t/, $row);
- push @rows, {'unmapped_reason_id' => $a[0],
- 'summary_description' => $a[1],
- 'full_description' => $a[2]};
-}
-$fh->close();
-
-foreach my $dbname (@dbnames) {
- print STDERR "updating $dbname\n";
- $db->do("use $dbname");
- my $sth = $db->prepare('DELETE FROM unmapped_reason');
- $sth->execute();
- $sth->finish();
-
- $sth = $db->prepare('INSERT INTO unmapped_reason (unmapped_reason_id,summary_description, full_description)
- VALUES (?,?,?)');
-
- foreach my $row (@rows) {
- print $row->{'unmapped_reason_id'}."\n";
- $sth->execute($row->{'unmapped_reason_id'},
- $row->{'summary_description'},
- $row->{'full_description'});
- }
-
- $sth->finish();
-}
-
-print STDERR "updates complete\n";
-
-
-sub usage {
- print STDERR <<EOF
-
- Usage: update_unmapped_reason_id options
- Where options are: -host hostname
- -user username
- -pass password
- -port port_of_server optional
- -release the release of the database to update used to
- match database names. e.g. 13
- -file the path of the file containing the insert statements
- of the entries of the external_db table
- -dbnames db1
- the names of the database to update. if not provided
- all of the core databases matching the release arg
- will be updated. Either -dbnames or -release must
- be specified, but not both. Multiple dbnames can
- be provided.
-
- E.g.:
-
- #update 2 databases
- perl update_unmapped_reasons.pl -host ecs1c -file unmapped_reason.txt -user ensadmin -pass secret -dbnames homo_sapiens_core_14_33 -dbnames mus_musculus_core_14_30
-
- #update all core databases for release 14
- perl update_unmapped_reasons.pl -host ecs2d -file unmapped_reason.txt -user ensadmin -pass secret -release 14
-
-EOF
-;
- exit;
-}
--
GitLab