tripal/tripal_analysis_interpro/tripal_analysis_interpro.module

<?php

require_once "parseInterpro.inc";

/*******************************************************************************
 * Tripal Interpro lets users show/hide iprscan results associated with a tripal
 * feature
 ******************************************************************************/
function tripal_analysis_interpro_init(){
	// Add javascript and style sheet
	drupal_add_css(drupal_get_path('theme', 'tripal').'/css/tripal_analysis_interpro.css');
	// Add javascript and style sheet
   drupal_add_js(drupal_get_path('theme', 'tripal').'/js/tripal_analysis_interpro.js');
}
/*******************************************************************************
 *  Provide information to drupal about the node types that we're creating
 *  in this module
 */
function tripal_analysis_interpro_node_info() {
	$nodes = array();
	$nodes['chado_analysis_interpro'] = array(
      'name' => t('Analysis: Interpro'),
      'module' => 'chado_analysis_interpro',
      'description' => t('An interpro analysis from the chado database'),
      'has_title' => FALSE,
      'title_label' => t('Analysis: Interpro'),
      'has_body' => FALSE,
      'body_label' => t('Interpro Analysis Description'),
      'locked' => TRUE
	);
	return $nodes;
}
/*******************************************************************************
*
*/
function tripal_analysis_interpro_block($op = 'list', $delta = 0, $edit=array()){
   switch($op) {
      case 'list':
         $blocks['tai_results']['info'] = t('Tripal InterProScan Analysis Results');
         $blocks['tai_results']['cache'] = BLOCK_NO_CACHE;
       return $blocks;

      case 'view':
         if(user_access('access chado_analysis_interpro content') and arg(0) == 'node' and is_numeric(arg(1))) {
            $nid = arg(1);
            $node = node_load($nid);
            $block = array();
            switch($delta){
               case 'tai_results':
                  $block['subject'] = t('InterProScan Results');
                  $block['content'] = theme('tripal_analysis_interpro_results', $node);
                  break;
                default :
            }
            return $block;
         }
   }
}
/*******************************************************************************
 *  Provide a Interpro Analysis form
 */
function chado_analysis_interpro_form ($node){

   // add in the default fields for the analysis
   $form = chado_analysis_form($node);

   $analysis = $node->analysis;

   $interprofile = $analysis->tripal_analysis_interpro->interprofile;
   $interproparameters = $analysis->tripal_analysis_interpro->interproparameters;
   $parsego = $analysis->tripal_analysis_interpro->parsego;
   $parseHTML = $analysis->tripal_analysis_interpro->parseHTML;
   $query_re = $analysis->tripal_analysis_interpro->query_re;
   $query_type = $analysis->tripal_analysis_interpro->query_type;
   $query_uniquename = $analysis->tripal_analysis_interpro->query_uniquename;


	$moreSettings ['interpro'] = 'Interpro Settings';
	$form['interpro'] = array(
      '#title' => t('Interpro Settings'),
      '#type' => 'fieldset',
      '#description' => t('Specific Settings for Interpro Analysis.'),
      '#collapsible' => TRUE,
      '#attributes' => array('id' => 'interpro-extra-settings'),
      '#weight' => 11
	);
	$form['interpro']['interprofile'] = array(
      '#title' => t('Interproscan Output File (in XML format)'),
      '#type' => 'textfield',
      '#description' => t('Please provide the full path to the XML output file generated by InterProScan.'),
      '#default_value' => $interprofile,
	);
	$form['interpro']['interprojob'] = array(
      '#type' => 'checkbox',
      '#title' => t('Submit a job to parse the InterProScan XML file'),
      '#description' => t('Note: features associated with the interpro results must '.
                             'exist in chado before parsing the file. Otherwise, interpro '.
                             'results that cannot be linked to a feature will be '.
                             'discarded.'),
      '#default_value' => $interprojob,
	   '#attributes' => array(
        'onclick' => 'return isSubmittingJob(this)'
      )
	);
	$form['interpro']['parseHTML'] = array(
      '#type' => 'checkbox',
      '#title' => t('The input file is in HTML format (deprecated, only provided for backwards compatibility)'),
      '#description' => t('Check the box to use the HTML parser. The feature name must be unique across all organisms and types.'),
      '#default_value' => $parseHTML
   );
	$form['interpro']['parsego'] = array(
      '#type' => 'checkbox',
      '#title' => t('Load GO terms to the database'),
      '#description' => t('Check the box to load GO terms to chado database'),
      '#default_value' => $parsego
   );
   $form['interpro']['interprokeywordjob'] = array(
      '#type' => 'checkbox',
      '#title' => t('Submit a job to extract keywords from the Interpro html output'),
      '#description' => t('Note: Interpro results are only searchable after keywords are extracted. Do not run this twice if you have already done so.'),
	);
	$form['interpro']['interproparameters'] = array(
      '#title' => t('Parameters'),
      '#type' => 'textfield',
      '#description' => t('The parameters used when running the InterProScan analysis.'),
      '#default_value' => $interproparameters,
	);

	$form['interpro']['query_re'] = array(
      '#title' => t('Query Name RE'),
      '#type' => 'textfield',
      '#description' => t('Enter the regular expression that will extract the '.
         'feature name from the query line in the interpro results. This should be '.
         'the same as the definition line in the query FASTA file.  This option is '.
         'is only required when the query does not identically match a feature '.
         'in the database.'),
      '#default_value' => $query_re,
	);

	$form['interpro']['query_type'] = array(
      '#title' => t('Query Type'),
      '#type' => 'textfield',
      '#description' => t('Please enter the Sequence Ontology term that describes '.
         'the query sequences used for InterProScan.  This is only necessary if two '.
         'or more sequences have the same name.'),
      '#default_value' => $query_type,
	);

	$form['interpro']['query_uniquename'] = array(
      '#title' => t('Use Unique Name'),
      '#type' => 'checkbox',
      '#description' => t('Select this checboxk if the query name in the results file '.
        'matches the uniquename of the feature.  By default, the blast results will '.
        'mapped to the "name" of the feature.'),
      '#default_value' => $query_uniquename,
	);
	return $form;
}
/**
 * 
 * 
 */
function chado_analysis_interpro_load($node){

   // load the default set of analysis fields
	$additions = chado_analysis_load($node);

   // create some variables for easier lookup
   $analysis = $additions->analysis;
   $analysis_id = $analysis->analysis_id;

   $intepro_settings  = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_settings');	
   $interprofile      = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_interprofile');
   $interproparameters= tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_interproparameters');
   $parsego           = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_parsego');		
   $parseHTML         = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_parseHTML');		
   $query_re          = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_query_re');	
   $query_type        = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_query_type');	
   $query_uniquename  = tripal_analysis_get_property($analysis->analysis_id,'analysis_interpro_query_uniquename');	

   $analysis->tripal_analysis_interpro->interprofile      = $interprofile->value;
   $analysis->tripal_analysis_interpro->interproparameters= $interproparameters->value;
   $analysis->tripal_analysis_interpro->parsego           = $parsego->value;
   $analysis->tripal_analysis_interpro->parseHTML         = $parseHTML->value;
   $analysis->tripal_analysis_interpro->query_re          = $query_re->value;
   $analysis->tripal_analysis_interpro->query_type        = $query_type->value;
   $analysis->tripal_analysis_interpro->query_uniquename  = $query_uniquename->value;

   // if there is an old style 'interpro_settings' array, then break these out for
   // use in the new format
   if(count($interpro_settings)>0){
	   $prop_values = explode ("|", $interpro_settings->value);
      $analysis->tripal_analysis_interpro->interprofile       = $prop_values[0];
      $analysis->tripal_analysis_interpro->interproparameters = $prop_values[1];
   }
	return $additions;
}
/**
 *  
 */
function chado_analysis_interpro_insert($node){
  // insert the analysistripal_core_generate_chado_var
   chado_analysis_insert($node);

   // set the type for this analysis
   tripal_analysis_insert_property($node->analysis_id,'analysis_type','tripal_analysis_interpro');

   // now add in the remaining settings as a single property but separated by bars
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_interprofile',$node->interprofile);
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_interproparameters',$node->interproparameters);
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_parsego',$node->parsego);		
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_parseHTML',$node->parseHTML);		
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_query_re',$node->query_re);	
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_query_type',$node->query_type);	
   tripal_analysis_insert_property($node->analysis_id,'analysis_interpro_query_uniquename',$node->query_uniquename);	
  
   // submit the parsing jobs
   chado_analysis_interpro_submit_job($node);

}
/**
 * 
 */
function chado_analysis_interpro_update($node){
  // insert the analysistripal_core_generate_chado_var
   chado_analysis_update($node);

   // set the type for this analysis
   tripal_analysis_update_property($node->analysis_id,'analysis_type','tripal_analysis_interpro',1);

   // now add in the remaining settings as a single property but separated by bars
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_interprofile',$node->interprofile,1);
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_interproparameters',$node->interproparameters,1);
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_parsego',$node->parsego,1);		
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_parseHTML',$node->parseHTML,1);		
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_query_re',$node->query_re,1);	
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_query_type',$node->query_type,1);	
   tripal_analysis_update_property($node->analysis_id,'analysis_interpro_query_uniquename',$node->query_uniquename,1);	

   // if this analysis uses the old style settings cvterm then remove that term
   $old = tripal_analysis_get_property($node->analysis_id,'analysis_interpro_settings');
   if(count($old) > 0){
      tripal_analysis_delete_property($node->analysis_id,'analysis_interpro_settings');
   }
  
   // submit the parsing jobs
   chado_analysis_interpro_submit_job($node);
}
/*******************************************************************************
 * Delete interpro anlysis
 */
function chado_analysis_interpro_delete($node){
	chado_analysis_delete($node);
}

/**
*
*/
function chado_analysis_interpro_submit_job($node){
   global $user;

	// Add a job if the user wants to parse the html output
	if($node->interprojob) {
		$job_args[0] = $node->analysis_id;
		$job_args[1] = $node->interprofile;
		if ($node->parsego) {
		   $job_args[2] = 1;
		} else {
         $job_args[2] = 0;
      }


		if (is_readable($node->interprofile)) {
			$fname = preg_replace("/.*\/(.*)/", "$1", $node->interprofile);
         if($node->parseHTML){
			   tripal_add_job("Parse HTML interpro: $fname",'tripal_analysis_interpro',
                           'tripal_analysis_interpro_parseHTMLFile', $job_args, $user->uid);
         } else {
		      $job_args[3] = $node->query_re;
		      $job_args[4] = $node->query_type;
		      $job_args[5] = $node->query_uniquename;;
			   tripal_add_job("Parse XML interpro: $fname",'tripal_analysis_interpro',
                           'tripal_analysis_interpro_parseXMLFile', $job_args, $user->uid);
         }
		} else {
			drupal_set_message("Can not open interpro output file. Job not scheduled.");
		}
	}
	
	// Add a job if the user wants to the keywords from the HTML output
	if ($node->interprokeywordjob) {
		$analysis_id = chado_get_id_for_node('analysis', $node);
		$job_args[0] = $analysis_id;
		tripal_add_job("Extract keywords for search: $node->analysisname",'tripal_analysis_interpro',
                           'tripal_analysis_interpro_extract_keywords', $job_args, $user->uid);
	}
}


/*******************************************************************************
 *  This function customizes the view of the chado_analysis node.  It allows
 *  us to generate the markup.
 */
function chado_analysis_interpro_view ($node, $teaser = FALSE, $page = FALSE) {
	// use drupal's default node view:
	if (!$teaser) {
		$node = node_prepare($node, $teaser);
		// When previewing a node submitting form, it shows 'Array' instead of
		// correct date format. We need to format the date here
		$time = $node->timeexecuted;
		if(is_array($time)){
			$month = $time['month'];
			$day = $time['day'];
			$year = $time['year'];
			$timestamp = $year.'-'.$month.'-'.$day;
			$node->timeexecuted = $timestamp;
		}
		// When viewing a node, we need to reformat the analysisprop since we
		// separate each value with a bar |
		if (preg_match("/.*\|.*/",$node->interprofile)) {
			$prop_values = explode("|", $node->interprofile);
			$node->interprofile = $prop_values[0];
			$node->interproparameters = $prop_values[1];
		}
	}
	return $node;
}


/*******************************************************************************
 * tripal_analysis_interpro_nodeapi()
 * HOOK: Implementation of hook_nodeapi()
 * Display interpro results for allowed node types
 */
function tripal_analysis_interpro_nodeapi(&$node, $op, $teaser, $page) {

	switch ($op) {
		case 'view':
			// Find out which node types for showing the interpro
			$types_to_show = variable_get('tripal_analysis_interpro_setting',
			array('chado_feature'));

			// Abort if this node is not one of the types we should show.
			if (!in_array($node->type, $types_to_show, TRUE)) {
				break;
			}
			// Add interpro to the content item if it's not a teaser
			if (!$teaser && $node->feature->feature_id) {
				if($node->build_mode == NODE_BUILD_SEARCH_INDEX){
					$node->content['tripal_analysis_interpro_index_version'] = array(
                  '#value' => theme('parse_EBI_Interpro_XML_index_version',$node),
                  '#weight' => 7,
					);
				} else {
					// Show interpro result if not at teaser view
					$node->content['tripal_feature_interpro_results'] = array(
                  '#value' => theme('tripal_feature_interpro_results', $node),
                  '#weight' => 7
					);
				}
			}
	}
}
/************************************************************************
 *  We need to let drupal know about our theme functions and their arguments.
 *  We create theme functions to allow users of the module to customize the
 *  look and feel of the output generated in this module
 */
function tripal_analysis_interpro_theme () {
	return array(
      'parse_EBI_Interpro_XML_index_version' => array (
         'arguments' => array('node'),
	   ),
      'tripal_feature_interpro_results' => array (
         'arguments' => array('node'=> null),
         'template' => 'tripal_feature_interpro_results',
	   )
	);
}
/*******************************************************************************
 * Prepare interpro result for the feature shown on the page
 */
//function theme_tripal_analysis_interpro_results ($node) {
//	$feature = $node->feature;
//	$content = tripal_get_interpro_results($feature->feature_id);
//	return $content;
//}

function tripal_analysis_interpro_preprocess_tripal_feature_interpro_results(&$variables){
   $feature = $variables['node']->feature;

   // we don't know how many analysis are the old style HTML version and how many are the newer
   // XML format.  So, to be backwards compatible we should get both.
   $variables['tripal_analysis_interpro']['results']['html'] = tripal_get_interpro_HTML_results($feature->feature_id);
	$variables['tripal_analysis_interpro']['results']['xml'] = tripal_get_interpro_XML_results($feature->feature_id);
}

/*******************************************************************************
 * Prepare interpro result for the feature shown on the page
 */
function theme_parse_EBI_Interpro_XML_index_version ($node) {
	$feature = $node->feature;
	$content = tripal_get_interpro_results_index_version($feature->feature_id);
	return $content;
}
/**
 * 
 */
function tripal_get_interpro_XML_results($feature_id){

	// Get the blast results stored as XML from the analysisfeatureprop table
   // the type for the property is named 'analysis_blast_output_iteration_hits'
   // and is found in the 'tripal' controlled vocabulary.  This CV term was
   // added by this module.
   $select = array(
      'analysisfeature_id' => array(
         'feature_id' => $feature_id,
      ), 
      'type_id' => array(
         'name' => 'analysis_interpro_xmloutput_hit',
         'cv_id' => array(
            'name' => 'tripal'
         ),
      ),         
   );
   $afeatureprops = tripal_core_chado_select('analysisfeatureprop',array('*'),$select);

   // iterate through all of the interpro XML properties for this feature
	$results = array ();
   if($afeatureprops){
	   foreach ($afeatureprops as $index => $afeatureprop) {

         // get the analysis feature record
         $analysisfeature_arr = tripal_core_chado_select('analysisfeature',array('analysis_id'),
            array('analysisfeature_id' => $afeatureprop->analysisfeature_id));
         $analysisfeature = $analysisfeature_arr[0];

         // get the analysis record and the analysis_id
         $analysis = tripal_core_generate_chado_var('analysis',
            array('analysis_id' => $analysisfeature->analysis_id));
         $analysis_id = $analysis->analysis_id;

         // parse the XML and add it to the array of blast results to be returned
         $interpro_xml = $afeatureprop->value;
         $orf = tripal_analysis_interpro_get_result_object($interpro_xml,$feature_id);
		   $results[$analysis->analysis_id]['protein_ORFs'][] = $orf;
         $results[$analysis->analysis_id]['analysis'] = $analysis;

         // iterate through all of the protein ORFs and combine the terms into one large list
         if(!is_array($results[$analysis->analysis_id]['allterms'])){
           $results[$analysis->analysis_id]['allterms'] = array();
         }
         if(!is_array($results[$analysis->analysis_id]['goterms'])){
           $results[$analysis->analysis_id]['goterms'] = array();
         }
         $results[$analysis->analysis_id]['allterms'] = array_merge($results[$analysis->analysis_id]['allterms'],$orf['iprterms']);
         $results[$analysis->analysis_id]['goterms'] = array_merge($results[$analysis->analysis_id]['goterms'],$orf['goterms']);
         $i++;
	   }
   }
   return $results;
}
/*******************************************************************************
 * tripal_get_interpro_results()
 * Get interpro result from featureprop table for the feature
 */
function tripal_get_interpro_HTML_results($feature_id){
	// Get cvterm_id for 'analysis_interpro_output_hit' which is required
	// for inserting into the analysisfeatureprop table
	$previous_db = tripal_db_set_active('chado');
	$sql = "SELECT CVT.cvterm_id FROM {cvterm} CVT 
	        INNER JOIN cv ON cv.cv_id = CVT.cv_id
	        WHERE CVT.name = 'analysis_interpro_output_hit'
	        AND CV.name = 'tripal'";
	$type_id = db_result(db_query($sql));
   
   // Get analysis times for the feature
	$sql = "SELECT A.analysis_id AS aid 
	        FROM {analysis} A
	        INNER JOIN analysisfeature AF ON A.analysis_id = AF.analysis_id 
	        INNER JOIN analysisfeatureprop AFP ON AF.analysisfeature_id = AFP.analysisfeature_id
	        WHERE feature_id = %d
	        AND AFP.type_id = %d
	        GROUP BY A.analysis_id
	       ";
   $hasResult = db_result(db_query($sql, $feature_id, $type_id));
   $result = db_query($sql, $feature_id, $type_id);

   // Show interpro result ORDER BY time
   if ($hasResult) { // If there is any result, show expandable box
      $content .= "<table class=\"tripal_interpro_results_table\">
                         <tr><td>";

      while ($ana = db_fetch_object($result)) {
         // Show analysis date
         $sql = "SELECT name, to_char(timeexecuted, 'MM-DD-YYYY') AS time 
         		  FROM {analysis} 
         		  WHERE analysis_id = %d";
         $ana_details = db_fetch_object(db_query($sql, $ana->aid));
         // Find node id for the analysis
         tripal_db_set_active($previous_db);
         $ana_nid = db_result(db_query("SELECT nid FROM {chado_analysis} WHERE analysis_id = %d", $ana->aid));
         $ana_url = url("node/".$ana_nid);
         $previous_db = tripal_db_set_active('chado');
       
         // Show content
         $content .= "<strong>Analysis Date:</strong> $ana_details->time
                        (<a href=$ana_url>$ana_details->name</a>)";
         
         // Show interpro results
         $sql = "SELECT AFP.value AS afpvalue
                 FROM {analysisfeatureprop} AFP
                 INNER JOIN analysisfeature AF on AF.analysisfeature_id = AFP.analysisfeature_id
                 WHERE AF.analysis_id = %d
                 AND AF.feature_id = %d
                ";
         $interpro_results = db_query($sql, $ana->aid, $feature_id);
      	 while ($afp = db_fetch_object($interpro_results)) {
      	 	$content .= $afp->afpvalue;
      	 }
      }   
      $content .= '</td></tr></table>';
   }
     
   tripal_db_set_active($previous_db);
	return $content;
}

/*******************************************************************************
 * tripal_get_interpro_results()
 * Get interpro result from featureprop table for the feature
 */
function tripal_get_interpro_results_index_version($feature_id){
   // Get cvterm_id for 'analysis_interpro_output_hit' which is required
   // for inserting into the analysisfeatureprop table
   $previous_db = tripal_db_set_active('chado');
   $sql = "SELECT CVT.cvterm_id FROM {cvterm} CVT ".
          "INNER JOIN cv ON cv.cv_id = CVT.cv_id ".
          "WHERE CVT.name = 'analysis_interpro_output_hit' ".
          "AND CV.name = 'tripal'";
   $type_id = db_result(db_query($sql));

   // Get xml string from analysisfeatureprop value column
   $sql = "SELECT AFP.value AS afpvalue FROM {analysisfeatureprop} AFP ".
          "INNER JOIN analysisfeature AF ON AF.analysisfeature_id = AFP.analysisfeature_id ".
          "INNER JOIN analysisprop AP ON AP.analysis_id = AF.analysis_id ".
          "WHERE feature_id = %d ".
          "AND AFP.type_id = %d ";
   $result = db_query($sql, $feature_id, $type_id);
   tripal_db_set_active($previous_db);
   if (db_result($result)) {
      // get the HTML content for viewing each of the XML file
      while ($analysisfeatureprop = db_fetch_object($result)) {
         $content .= $analysisfeatureprop->afpvalue;
      }
   }
   return $content;
}

/*******************************************************************************
 * Tripal Interpro administrative setting form. This function is called by
 * tripal_analysis module which asks for an admin form to show on the page
 */
function tripal_analysis_interpro_get_settings() {
	// Get an array of node types with internal names as keys
	$options = node_get_types('names');
	// Add 'chado_feature' to allowed content types for showing interpro results
	$allowedoptions ['chado_feature'] = "Show Interpro results on feature pages";

	$form['description'] = array(
       '#type' => 'item',
       '#value' => t("Some chado features were analyzed using InterProScan. This option allows user to display the interpro analysis results. Please read user manual for storage and display of interpro files. Check the box to enable the analysis results. Uncheck to disable it."),
       '#weight' => 0,
	);

	$form['tripal_analysis_interpro_setting'] = array(
      '#type' => 'checkboxes',
      '#options' => $allowedoptions,
      '#default_value' => variable_get('tripal_analysis_interpro_setting',
	   array('chado_feature')),
	);

	$settings->form = $form;
	$settings->title = "Tripal Interpro";
	return $settings;
}

/*******************************************************************************
 * Set the permission types that the chado module uses.  Essentially we
 * want permissionis that protect creation, editing and deleting of chado
 * data objects
 */
function tripal_analysis_interpro_perm(){
   return array(
      'access chado_analysis_interpro content',
      'create chado_analysis_interpro content',
      'delete chado_analysis_interpro content',
      'edit chado_analysis_interpro content',
   );
}

/*******************************************************************************
 *  The following function proves access control for users trying to
 *  perform actions on data managed by this module
 */
function chado_analysis_interpro_access($op, $node, $account){
   if ($op == 'create') {
      return user_access('create chado_analysis_interpro content', $account);
   }

   if ($op == 'update') {
      if (user_access('edit chado_analysis_interpro content', $account)) {
         return TRUE;
      }
   }
   if ($op == 'delete') {
      if (user_access('delete chado_analysis_interpro content', $account)) {
         return TRUE;
      }
   }
   if ($op == 'view') {
      if (user_access('access chado_analysis_interpro content', $account)) {
         return TRUE;
      }
   }
   return FALSE;
}

/*******************************************************************************
 * Parsing Interpro HTML results that are stored in analysisfeatureprop for 
 * searching 
 * */
function tripal_analysis_interpro_extract_keywords ($analysis_id) {	
	
	print "Extracting keywords...\n";
	// Get all interpro output hits except for records with 'No hits reported', 'parent', 'children'.
	$output_type_id = tripal_get_cvterm_id('analysis_interpro_output_hit');
	$sql = "SELECT AFP.analysisfeature_id, AFP.value FROM {analysisfeatureprop} AFP 
					 INNER JOIN {analysisfeature} AF ON AF.analysisfeature_id = AFP.analysisfeature_id
					 WHERE type_id = $output_type_id 
					 AND AF.analysis_id = $analysis_id
					 AND value NOT like '%No hits reported.%' 
           		 AND value NOT like '%parent%' 
           		 AND value NOT like '%children%'";
	$results = chado_query($sql);
	$keyword_type_id = tripal_get_cvterm_id('analysis_interpro_output_keywords');
	// Define what to be extracted in the array
	$search = array (
							"'SEQUENCE:.*'",
							"'CRC64:.*'",
							"'LENGTH:.*'",
							"'<b>InterPro<br/>'",
							"'<br/>Domain|Family\n'",
							"'<td>no description</td>'",
                 		"'<[/!]*?[^<>]*?>'si",          // replace HTML tags with a space
							"'\n'", // replace newlines with a space
	);
	$replace = array (
								"",
								"",
								"",
								"",
								"", 
								"",
					  			" ",
                 			" ",
	);
	while ($record = db_fetch_object($results)) {
		$af_id = $record->analysisfeature_id;
		$value = $record->value;

		// Retrive keywords for this analysisfeature_id if there is any
		$sql = "SELECT value FROM {analysisfeatureprop} WHERE analysisfeature_id =$af_id AND type_id = $keyword_type_id";
		$keywords = db_result(chado_query($sql));
		
		// Extract new keywords from the interpro html output
		$text = preg_replace($search, $replace, $value); 
		$new_keywords = trim(ereg_replace(' +', ' ', $text)); // remove extra spaces
		
		// Append the new keywords 
		if ($keywords) {
			$new_keywords = "$keywords $new_keywords";
			$sql = "UPDATE {analysisfeatureprop} SET value = '$new_keywords' WHERE analysisfeature_id =$af_id AND type_id = $keyword_type_id ";
		} else {
			// Insert the keyword into the analysisfeatureprop table
			$sql = "INSERT INTO {analysisfeatureprop} (analysisfeature_id, type_id, value, rank) VALUES ($af_id, $keyword_type_id, '$new_keywords', 0)";
		}
		chado_query($sql);

	}
	
	print "Finished.\n";
	
}