tripal_analysis_kegg.install

<?php

/**
 *  Implementation of hook_install();
 */
function tripal_analysis_kegg_install() {

  // Register this analysis module.
  tripal_register_analysis_child('tripal_analysis_kegg');

  // Add the cvterms needed for this module.
  tripal_analysis_kegg_add_cvterms();

  // Make sure the KEGG databases we will be using exist.  The list of
  // KEGG databases can be found here: http://www.genome.jp/kegg/kegg3.html
  // we only use three: KEGG PATHWAY, KEGG BRITE AND KEGG ORTHOLOGY.
  tripal_insert_db(array(
   'name' => 'KEGG_PATHWAY',
   'description' => 'KEGG PATHWAY is a collection of manually ' .
       'drawn pathway maps representing our knowledge on the molecular interaction ' .
       'and reaction networks.',
   'url' => 'http://www.genome.jp/kegg/pathway.html',
   'urlprefix' =>  'http://www.genome.jp/dbget-bin/www_bget?ko+'
  ));
  tripal_insert_db(array(
   'name' => 'KEGG_BRITE',
   'description' => 'KEGG BRITE is a collection of hierarchical ' .
       'classifications from KEGG.',
   'url' => 'http://www.genome.jp/kegg/brite.html',
   'urlprefix' =>   'http://www.genome.jp/kegg-bin/get_htext?'
  ));
  tripal_insert_db(array(
   'name' => 'KEGG_ORTHOLOGY',
   'description' => 'The KEGG Orthology (KO) system ' .
       'consists of manually defined ortholog groups that correspond to ' .
      'pathways, BRITE hierarchys and modules',
   'url' => 'http://www.genome.jp/kegg/ko.html',
   'urlprefix' =>  'http://www.genome.jp/dbget-bin/www_bget?ko:'
  ));
  tripal_insert_db(array(
   'name' => 'KEGG_MODULE',
   'description' => 'KEGG MODULE is a collection of manually '.
       'defined functional units, called KEGG modules and identified by the '.
       'M numbers, used for annotation and biological interpretation of '.
       'sequenced genomes',
   'url' => 'http://www.genome.jp/kegg/module.html',
   'urlprefix' => 'http://www.genome.jp/kegg-bin/show_module?',
  ));

  // Add in the controlled vocabularies for the KEGG database.
  tripal_insert_cv('KEGG_PATHWAY','Holds terms from the KEGG PATHWAY database.');
  tripal_insert_cv('KEGG_BRITE', 'Holds terms from the KEGG BRITE database.');
  tripal_insert_cv('KEGG_ORTHOLOGY', 'Holds terms from the KEGG ORTHOLOGY database.');
  tripal_insert_cv('KEGG_MODULE',  'Holds terms from the KEGG MODULE database.');

  // Add the materialized view
  tripal_analysis_kegg_add_mview_kegg_by_organism();
}


/**
 * Creates a view showing the kegg results for each organism.
 *
 * @ingroup tripal_analysis
 */
function tripal_analysis_kegg_add_mview_kegg_by_organism() {
  $view_name = 'kegg_by_organism';
  $comment = t('This view is for associating an organism to a KEGG analysis.');

  // this is the SQL used to identify the organism to which an analsysis
  // has been used.  This is obtained though the analysisfeature -> feature -> organism
  // joins
  $sql = "
    SELECT DISTINCT A.name,A.analysis_id,F.organism_id
    FROM analysisprop AP
      INNER JOIN analysis A on A.analysis_id = AP.analysis_id
      INNER JOIN cvterm CVT on CVT.cvterm_id = AP.type_id
      INNER JOIN analysisfeature AF on AF.analysis_id = A.analysis_id
      INNER JOIN feature F on F.feature_id = AF.feature_id
    WHERE CVT.name = 'analysis_kegg_settings'
  ";

  // the schema array for describing this view
  $schema = array(
    'table' => $view_name,
    'description' => $comment,
    'fields' => array(
      'analysis_name' => array(
        'type' => 'varchar',
        'length' => 255,
      ),
      'analysis_id' => array(
        'type' => 'int',
        'not null' => TRUE,
      ),
      'organism_id' => array(
        'type' => 'int',
        'not null' => TRUE,
      ),
    ),
    'indexes' => array(
      'kegg_by_organism_indx0' => array('analysis_id'),
      'kegg_by_organism_indx1' => array('organism_id'),
    ),
    'foreign keys' => array(
      'analysis' => array(
        'table' => 'analysis',
        'columns' => array(
          'analysis_id' => 'analysis_id',
        ),
      ),
      'organism' => array(
        'table' => 'organism',
        'columns' => array(
          'organism_id' => 'organism_id',
        ),
      ),
    ),
  );

  // add the view
  tripal_add_mview($view_name, 'tripal_analysis_kegg', $schema, $sql, $comment);
}

/**
*
*/
function tripal_analysis_kegg_add_cvterms() {
  tripal_insert_cvterm(array(
    'name' => 'analysis_kegg_settings',
    'def' => 'Settings of a KEGG analysis, Currently include only the heir.tar.gz file name & path.',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));
  tripal_insert_cvterm(array(
    'name' => 'kegg_brite_data',
    'def' => 'This term is intended for use in the analysisfeatureprop table to store the KAAS results data.',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));
  tripal_insert_cvterm(array(
    'name' => 'analysis_kegg_query_re',
    'def' => 'The regular expression for finding the feature name in the query definition line of the blast results',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));
  tripal_insert_cvterm(array(
    'name' => 'analysis_kegg_query_type',
    'def' => 'The feature type (e.g. mRNA, polypeptide) of the query input file. This type is used to identify the query feature when multiple features have the same name',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));
  tripal_insert_cvterm(array(
    'name' => 'analysis_kegg_query_uniquename',
    'def' => 'Indicates if the matched name in the query definition line of the blast results is feature uniquename',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));
  tripal_insert_cvterm(array(
    'name' => 'analysis_kegg_output_keywords',
    'def' => 'Selected keywords for kegg html output which are indexed for search.',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));
  tripal_insert_cvterm(array(
    'name' => 'analysis_kegg_organism_id',
    'def' => 'The organism ID of the organism to which KEGG analysis results belong.',
    'cv_name' => 'tripal',
    'db_name' => 'tripal',
  ));

}
/*******************************************************************************
* Implementation of hook_uninstall()
*/
function tripal_analysis_kegg_uninstall() {
  tripal_unregister_analysis_child('tripal_analysis_kegg');

  $mview_id = tripal_get_mview_id('kegg_by_organism');
  if ($mview_id) {
    tripal_delete_mview($mview_id);
  }
}

/*******************************************************************************
 * Implementation of hook_requirements(). Make sure 'Tripal Core' and 'Tripal
 * Analysis' are enabled before installation
 */
function tripal_analysis_kegg_requirements($phase) {
  $requirements = array();
  if ($phase == 'install') {
    // make sure chado is installed
    if (!$GLOBALS["chado_is_installed"]) {
      $requirements ['tripal_feature'] = array(
          'title' => "t ripal_feature",
          'value' => "ERROR: Chado must be installed before this module can be enabled",
          'severity' => REQUIREMENT_ERROR,
      );
    }
  }
  return $requirements;
}

/**
 * Fixes an error with the materialized view installation
 *
 */
function tripal_analysis_kegg_update_7200() {

  // there is a bug in the Tripal v2.0-alpha release that didn't add the
  // materialized view schema to the mviews table.
  // get the schema for the materialized view from the custom_tables table
  // as there is a copy there, but only if the schema is missing from the
  // materialized view table
  try {
    $view_name = 'kegg_by_organism';
    $schema = db_select('tripal_mviews', 'tm')
      ->fields('tm', array('mv_schema'))
      ->condition('name', $view_name)
      ->execute()
      ->fetchField();
    if (!$schema or $schema == 'Array') {
      $schema = db_select('tripal_custom_tables', 'tct')
        ->fields('tct', array('schema'))
        ->condition('table_name', $view_name)
        ->execute()
        ->fetchField();
      $schema_str = var_export(unserialize($schema), TRUE);
      $schema_str = preg_replace('/=>\s+\n\s+array/', '=> array', $schema_str);
      db_update('tripal_mviews')
        ->fields(array('mv_schema' => $schema_str))
        ->condition('name', $view_name)
        ->execute();
    }
  }
  catch (\PDOException $e) {
    $error = $e->getMessage();
    throw new DrupalUpdateException('Failed to complete update' . $error);
  }
}

/**
 * Adds an organism_id property for KEGG Analysis results
 *
 */
function tripal_analysis_kegg_update_7201() {
  try {
    tripal_insert_cvterm(array(
      'name' => 'analysis_kegg_organism_id',
      'definition' => 'The organism ID of the organism to which KEGG analysis results belong.',
      'cv_name' => 'tripal',
      'db_name' => 'tripal'
    ));
  }
  catch (\PDOException $e) {
    $error = $e->getMessage();
    throw new DrupalUpdateException('Failed to complete update: ' . $error);
  }
}

/**
 * Adds the KEGG Module database and vocabulary
 */
function tripal_analysis_kegg_update_7202() {
  try {
    tripal_insert_db(array(
      'name' => 'KEGG_MODULE',
      'description' => 'KEGG MODULE is a collection of manually '.
      'defined functional units, called KEGG modules and identified by the '.
      'M numbers, used for annotation and biological interpretation of '.
      'sequenced genomes',
      'url' => 'http://www.genome.jp/kegg/module.html',
      'urlprefix' => 'http://www.genome.jp/kegg-bin/show_module?',
    ));
    tripal_insert_cv('KEGG_MODULE',  'Holds terms from the KEGG MODULE database.');
  }
  catch (\PDOException $e) {
    $error = $e->getMessage();
    throw new DrupalUpdateException('Failed to complete update: ' . $error);
  }
}