dbolser

#!perl

use strict;
use warnings;

## Easy manipulation of sets of integers (arbitrary intervals)
use Set::IntRange;

## Create two sets to play with
my $set_len = 1000000000;

dbolser

<?php
/**
 * @file
 * Apache Solr external index module - example module for using external data.
 */

/**
 * implements hook_apachesolr_query_alter().
 */
function apachesolr_external_index_apachesolr_query_alter(DrupalSolrQueryInterface $query) {

dbolser

#!  perl

use strict;
use warnings;

die "pass an intron-style GFF for me to calcualte exons over\n"
  unless @ARGV;

## Tracking variables
my $prev_type = 'dumped';

dbolser

for module in \
    ensembl \
    ensembl-compara \
    ensembl-funcgen \
    ensembl-rest \
    ensembl-tools \
    ensembl-variation
do
    if [ ! -d "$module" ]; then
        echo "Checking out $module ($branch)"

dbolser

DROP TABLE IF EXISTS test_et;
CREATE TABLE         test_et (
  exon_id       int unsigned NOT NULL,
  transcript_id int unsigned NOT NULL,
  rank          int          NOT NULL,
  ##
  PRIMARY KEY    (exon_id, transcript_id, rank)#,
  #KEY exon       (exon_id),
  #KEY transcript (         transcript_id)
);

dbolser

> SELECT COUNT(*), COUNT(DISTINCT exon_id), COUNT(DISTINCT transcript_id) FROM exon_transcript;
+----------+-------------------------+-------------------------------+
| COUNT(*) | COUNT(DISTINCT exon_id) | COUNT(DISTINCT transcript_id) |
+----------+-------------------------+-------------------------------+
|   248445 |                  248445 |                        248445 |
+----------+-------------------------+-------------------------------+
1 row in set (0.00 sec)


> SELECT COUNT(*), SUM(exon_id is null), COUNT(DISTINCT exon_id), COUNT(DISTINCT transcript_id) FROM exon_transcript;

dbolser

[email protected]:4275 (oryza_sativa_core_22_75_7)
> SELECT COUNT(*) FROM exon_transcript;
+----------+
| COUNT(*) |
+----------+
|   248445 |
+----------+
1 row in set (0.00 sec)

dbolser

SELECT
  s.name
FROM
  seq_region s
INNER JOIN
  seq_region_attrib USING (seq_region_id)
INNER JOIN
  coord_system c USING (coord_system_id)
LEFT JOIN
  assembly ON seq_region_id = asm_seq_region_id

dbolser

SERVER1=mysql-devel-2
SERVER2=mysql-staging-1-ensrw

A="SELECT xref.* "
B="SELECT object_xref.* "
C="SELECT ontology_xref.* "

BASE="
FROM
  ontology_xref

dbolser

CURRENT:
solanum_lycopersicum_core_21_74_240
+----------+------+-------+---------------------+----------------+----------------+----------------------+
| COUNT(*) | XREF | OBJS  | ensembl_object_type | external_db_id | external_db_id | db_name              |
+----------+------+-------+---------------------+----------------+----------------+----------------------+
|    44538 | 1470 | 15954 | Translation         |           1000 |           1200 | Interpro             |
|     6082 |  710 |  1266 | Translation         |           1000 |           2000 | Uniprot/SPTREMBL     |
|     2286 |  426 |   333 | Translation         |           1000 |           2200 | Uniprot/SWISSPROT    |
|    37342 | 2956 |  6694 | Translation         |           1000 |          50720 | Ensembl Plants       |
+----------+------+-------+---------------------+----------------+----------------+----------------------+