tripal/tripal_analysis_interpro/tripal_analysis_interpro.module

<?php
// $Id:
// Copyright 2009 Clemson University
/*******************************************************************************
 * 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;
}

/*******************************************************************************
 *  Provide a Interpro Analysis form
 */
function chado_analysis_interpro_form ($node){

	$type = node_get_types('type', $node);
	$form = array();
	$form['title']= array(
      '#type' => 'hidden',
      '#default_value' => $node->title,
	);
	$form['analysisname']= array(
      '#type' => 'textfield',
      '#title' => t('Analysis Name'),
      '#required' => FALSE,
      '#default_value' => $node->analysisname,
      '#weight' => 1
	);
	$form['program']= array(
      '#type' => 'textfield',
      '#title' => t('Program'),
      '#required' => TRUE,
      '#default_value' => $node->program,
      '#weight' => 2
	);
	$form['programversion']= array(
      '#type' => 'textfield',
      '#title' => t('Program Version'),
      '#required' => TRUE,
      '#default_value' => $node->programversion,
      '#weight' => 3
	);
	$form['algorithm']= array(
      '#type' => 'textfield',
      '#title' => t('Algorithm'),
      '#required' => FALSE,
      '#default_value' => $node->algorithm,
      '#weight' => 4
	);
	$form['sourcename']= array(
      '#type' => 'textfield',
      '#title' => t('Source Name'),
      '#required' => FALSE,
      '#default_value' => $node->sourcename,
      '#weight' => 5
	);
	$form['sourceversion']= array(
      '#type' => 'textfield',
      '#title' => t('Source Version'),
      '#required' => FALSE,
      '#default_value' => $node->sourceversion,
      '#weight' => 6
	);
	$form['sourceuri']= array(
      '#type' => 'textfield',
      '#title' => t('Source URI'),
      '#required' => FALSE,
      '#default_value' => $node->sourceuri,
      '#weight' => 7
	);
	// Get time saved in chado
	$default_time = $node->timeexecuted;
	$year = preg_replace("/^(\d+)-\d+-\d+ .*/", "$1", $default_time);
	$month = preg_replace("/^\d+-0?(\d+)-\d+ .*/", "$1", $default_time);
	$day = preg_replace("/^\d+-\d+-0?(\d+) .*/", "$1", $default_time);
	// If the time is not set, use current time
	if (!$default_time) {
		$default_time = time();
		$year = format_date($default_time, 'custom', 'Y');
		$month = format_date($default_time, 'custom', 'n');
		$day = format_date($default_time, 'custom', 'j');
	}
	$form['timeexecuted']= array(
      '#type' => 'date',
      '#title' => t('Time Executed'),
      '#required' => TRUE,
      '#default_value' => array(
         'year' => $year,
         'month' => $month,
         'day' => $day,
	),
      '#weight' => 8
	);
	$form['description']= array(
      '#type' => 'textarea',
      '#rows' => 15,
      '#title' => t('Description and/or Program Settings'),
      '#required' => FALSE,
      '#default_value' => check_plain($node->description),
      '#weight' => 9
	);

	//----InterProScan Settings (Shown only when Tripal Interpro is enabled) ----
	if (preg_match("/.*\|.*/",$node->interprofile)) {
		$prop_values = explode("|", $node->interprofile);
		$node->interprofile = $prop_values[0];
		$node->interproparameters = $prop_values[1];
	}
	$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 html format)'),
      '#type' => 'textfield',
      '#description' => t('The html output file generated by Interproscan in full path.'),
      '#default_value' => $node->interprofile,
	);
	$form['interpro']['interprojob'] = array(
      '#type' => 'checkbox',
      '#title' => t('Submit a job to parse the Interpro html output'),
      '#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. Also, Triapl Interpro module needs to be enabled.'),
      '#default_value' => $node->interprojob,
	   '#attributes' => array(
        'onclick' => 'return isSubmittingJob(this)'
      )
	);
	$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' => $node->parsego
   );
	$form['interpro']['interproparameters'] = array(
      '#title' => t('Parameters'),
      '#type' => 'textfield',
      '#description' => t('The parameters for running the interpro analysis.'),
      '#default_value' => $node->interproparameters,
	);
	return $form;
}
/*******************************************************************************
 *  
 */
function chado_analysis_interpro_insert($node){
	global $user;
	// Create a timestamp so we can insert it into the chado database
	$time = $node->timeexecuted;
	$month = $time['month'];
	$day = $time['day'];
	$year = $time['year'];
	$timestamp = $month.'/'.$day.'/'.$year;

	// If this analysis already exists then don't recreate it in chado
	$analysis_id = $node->analysis_id;
	if ($analysis_id) {
		$sql = "SELECT analysis_id ".
               "FROM {Analysis} ".
               "WHERE analysis_id = %d ";
		$previous_db = tripal_db_set_active('chado');
		$analysis = db_fetch_object(db_query($sql, $node->analysis_id));
		tripal_db_set_active($previous_db);
	}

	// If the analysis doesn't exist then let's create it in chado.
	if(!$analysis){
		// First add the item to the chado analysis table
		$sql = "INSERT INTO {analysis} ".
             "  (name, description, program, programversion, algorithm, ".
             "   sourcename, sourceversion, sourceuri, timeexecuted) ".
             "VALUES ('%s','%s','%s','%s','%s','%s','%s','%s','%s')";
		$previous_db = tripal_db_set_active('chado');  // use chado database
		db_query($sql,$node->analysisname, $node->description,
		$node->program,$node->programversion,$node->algorithm,
		$node->sourcename, $node->sourceversion, $node->sourceuri,
		$timestamp);

		// find the newly entered analysis_id
		$sql = "SELECT analysis_id ".
             "FROM {Analysis} ".
             "WHERE program='%s'".
             "AND programversion='%s'".
             "AND sourcename='%s'";
		$analysis_id = db_result(db_query($sql, $node->program,
		$node->programversion, $node->sourcename));

		// Get cvterm_id for 'analysis_interpro_settings'
		$sql = "SELECT CVT.cvterm_id FROM {cvterm} CVT ".
             "INNER JOIN cv ON cv.cv_id = CVT.cv_id ".
             "WHERE CVT.name = 'analysis_interpro_settings' ".
             "AND CV.name = 'tripal'";
		$type_id = db_result(db_query($sql));

		// Insert into chado {analysisprop} table
		$sql = "INSERT INTO {analysisprop} (analysis_id, type_id, value) ".
             "VALUES (%d, %d, '%s')";
		$interprosettings = $node->interprofile."|".$node->interproparameters;
		db_query($sql, $analysis_id, $type_id, $interprosettings);

		tripal_db_set_active($previous_db);  // switch back to drupal database
		// Add a job if the user wants to parse the html output
		if($node->interprojob) {
			$job_args[0] = $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);
				tripal_add_job("Parse interpro: $fname",'tripal_analysis_interpro',
                           'tripal_analysis_interpro_parseHTMLFile', $job_args, $user->uid);
			} else {
				drupal_set_message("Can not open interpro output file. Job not scheduled.");
			}
		}
	}

	// Make sure the entry for this analysis doesn't already exist in the
	// chado_analysis table if it doesn't exist then we want to add it.
	$node_check_sql = "SELECT * FROM {chado_analysis} ".
                     "WHERE analysis_id = %d";
	$node_check = db_fetch_object(db_query($node_check_sql, $analysis_id));
	if(!$node_check){
		// next add the item to the drupal table
		$sql = "INSERT INTO {chado_analysis} (nid, vid, analysis_id) ".
             "VALUES (%d, %d, %d)";
		db_query($sql,$node->nid,$node->vid,$analysis_id);
		// Create a title for the analysis node using the unique keys so when the
		// node is saved, it will have a title
		$record = new stdClass();
		// If the analysis has a name, use it as the node title. If not, construct
		// the title using program, programversion, and sourcename
		if ($node->analysisname) {
			$record->title = $node->analysisname;
		} else {
			//Construct node title as "program (version)
			$record->title = "$node->program ($node->programversion)";
		}
		$record->nid = $node->nid;
		drupal_write_record('node',$record,'nid');
		drupal_write_record('node_revisions',$record,'nid');
	}
}
/*******************************************************************************
 * Delete interpro anlysis
 */
function chado_analysis_interpro_delete($node){
	// Before removing, get analysis_id so we can remove it from chado database
	// later
	$sql_drupal = "SELECT analysis_id ".
                 "FROM {chado_analysis} ".
                 "WHERE nid = %d ".
                 "AND vid = %d";
	$analysis_id = db_result(db_query($sql_drupal, $node->nid, $node->vid));

	// Remove data from the {chado_analysis}, {node}, and {node_revisions} tables
	$sql_del = "DELETE FROM {chado_analysis} ".
              "WHERE nid = %d ".
              "AND vid = %d";
	db_query($sql_del, $node->nid, $node->vid);
	$sql_del = "DELETE FROM {node} ".
              "WHERE nid = %d ".
              "AND vid = %d";
	db_query($sql_del, $node->nid, $node->vid);
	$sql_del = "DELETE FROM {node_revisions} ".
              "WHERE nid = %d ".
              "AND vid = %d";
	db_query($sql_del, $node->nid, $node->vid);

	//Remove from analysisfeatureprop, analysisfeature, analysis, and analysisprop tables
   $previous_db = tripal_db_set_active('chado');
   $sql = "SELECT analysisfeature_id FROM {analysisfeature} WHERE analysis_id=%d";
   $results = db_query($sql, $analysis_id);
   while ($af = db_fetch_object($results)) {
      db_query("DELETE FROM {analysisfeatureprop} WHERE analysisfeature_id = %d", $af->analysisfeature_id);
   }
   db_query("DELETE FROM {analysisfeature} WHERE analysis_id = %d", $analysis_id);
   db_query("DELETE FROM {analysisprop} WHERE analysis_id = %d", $analysis_id);
   db_query("DELETE FROM {analysis} WHERE analysis_id = %d", $analysis_id);
   tripal_db_set_active($previous_db);
}

/*******************************************************************************
 * Update interpro analysis
 */
function chado_analysis_interpro_update($node){
	global $user;
	if($node->revision){
		// TODO -- decide what to do about revisions
	} else {
		// Create a timestamp so we can insert it into the chado database
		$time = $node->timeexecuted;
		$month = $time['month'];
		$day = $time['day'];
		$year = $time['year'];
		$timestamp = $month.'/'.$day.'/'.$year;

		// get the analysis_id for this node:
		$sql = "SELECT analysis_id ".
             "FROM {chado_analysis} ".
             "WHERE vid = %d";
		$analysis_id = db_fetch_object(db_query($sql, $node->vid))->analysis_id;

		$sql = "UPDATE {analysis} ".
             "SET name = '%s', ".
             "    description = '%s', ".
             "    program = '%s', ".
             "    programversion = '%s', ".
             "    algorithm = '%s', ".
             "    sourcename = '%s', ".
             "    sourceversion = '%s', ".
             "    sourceuri = '%s', ".
             "    timeexecuted = '%s' ".
             "WHERE analysis_id = %d ";

		$previous_db = tripal_db_set_active('chado');  // use chado database
		db_query($sql, $node->analysisname, $node->description, $node->program,
		$node->programversion,$node->algorithm,$node->sourcename,
		$node->sourceversion, $node->sourceuri, $timestamp, $analysis_id);

		// Get cvterm_id for 'analysis_interpro_settings'
		$sql = "SELECT CVT.cvterm_id FROM {cvterm} CVT ".
             "INNER JOIN cv CV ON CV.cv_id = CVT.cv_id ".
             "WHERE CVT.name = 'analysis_interpro_settings' ".
             "AND CV.name = 'tripal'";
		$type_id = db_result(db_query($sql));
		 
		$sql = "UPDATE {analysisprop} ".
             "SET value = '%s' ".
             "WHERE analysis_id = %d AND type_id = %d";
		$interprosettings = $node->interprofile."|".$node->interproparameters;
		db_query($sql, $interprosettings, $analysis_id, $type_id);
		tripal_db_set_active($previous_db);  // switch back to drupal database
		// Add a job if the user wants to parse the html output
		if($node->interprojob) {
			$job_args[0] = $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);
				tripal_add_job("Parse interpro: $fname",'tripal_analysis_interpro',
                           'tripal_analysis_interpro_parseHTMLFile', $job_args, $user->uid);
			} else {
				drupal_set_message("Can not open interpro output file. Job not scheduled.");
			}
		}

		// Create a title for the analysis node using the unique keys so when the
		// node is saved, it will have a title
		$record = new stdClass();
		// If the analysis has a name, use it as the node title. If not, construct
		// the title using program, programversion, and sourcename
		if ($node->analysisname) {
			$record->title = $node->analysisname;
		} else {
			//Construct node title as "program (version)
			$record->title = "$node->program ($node->programversion)";
		}
		$record->nid = $node->nid;
		drupal_write_record('node',$record,'nid');
		drupal_write_record('node_revisions',$record,'nid');
	}
}
/*******************************************************************************
 *  When a node is requested by the user this function is called to allow us
 *  to add auxiliary data to the node object.
 */
function chado_analysis_interpro_load($node){
	// get the analysis_id for this node:
	$sql = "SELECT analysis_id FROM {chado_analysis} WHERE vid = %d";
	$ana_node = db_fetch_object(db_query($sql, $node->vid));
	$additions = new stdClass();
	if ($ana_node) {
		// get analysis information
		$sql = "SELECT Analysis_id, name AS analysisname, description, program, ".
             "  programversion, algorithm, sourcename, sourceversion, ".
             "  sourceuri, timeexecuted ".
             "FROM {Analysis} ".
             "WHERE Analysis_id = $ana_node->analysis_id";
		$previous_db = tripal_db_set_active('chado');  // use chado database
		$additions = db_fetch_object(db_query($sql));

		// get cvterm_id for 'analysis_interpro_settings'
		$sql = "SELECT CVT.cvterm_id FROM {cvterm} CVT ".
             "INNER JOIN cv ON cv.cv_id = CVT.cv_id ".
             "WHERE CVT.name = 'analysis_interpro_settings' ".
             "AND CV.name = 'tripal'";
		$type_id = db_result(db_query($sql));
		// get analysisprop information
		$sql = "SELECT value FROM {analysisprop} ".
             "WHERE analysis_id = %d ".
             "AND type_id = %d";
		$analysisprop = db_result(db_query($sql, $ana_node->analysis_id, $type_id));
		$prop_values = explode ("|", $analysisprop, 1);
		$additions->interprofile = $prop_values[0];
		$additions->interproparameters = $prop_values[1];

		tripal_db_set_active($previous_db);  // now use drupal database
	}
	// If the analysis has a name, use it as the node title. If not, construct
	// the title using program programversion, and sourcename
	if ($additions->analysisname) {
		$additions->title = $additions->analysisname;
	} else {
		// Construct node title as "program version (source)
		$additions->title = "$additions->program ($additions->programversion)";
	}
	return $additions;
}

/*******************************************************************************
 *  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;
}

/*******************************************************************************
 * Parse Interpro HTML Output file into analysisfeatureprop table
 */
function tripal_analysis_interpro_parseHTMLFile ($analysis_id, $interprofile, $parsego, $job_id) {

	// Prepare log
	$filename = preg_replace("/.*\/(.*)/", "$1", $interprofile);
	$logfile = file_directory_path() . "/tripal/tripal_analysis_interpro/load_$filename.log";
	$log = fopen($logfile, 'a'); // append parsing results to log file

	// Parsing started
	print "Parsing File:".$interprofile." ...\n";
	fwrite($log, date("D M j G:i:s Y").". Loading $interprofile\n");

	// Get cvterm_id for 'analysis_interpro_output_iteration_hits' which is required
	// for inserting into the analysisfeatureprop table
	$previous_db = tripal_db_set_active('chado'); // use chado database
	$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));
	print "cvterm_id for analysis_interpro_output_iteration_hits is $type_id\n";

	// Load the HTML file and convert it into XML for loading
	$dom = new domDocument;
	$dom->loadHTMLFile($interprofile);
	$xml = $dom->saveXML();
	$interproput = simplexml_load_string($xml);

	// Get html tables for parsing
	$tables = $interproput->children()->children();

	// Count the number of tables to be processed
	$no_iterations = 0;
	foreach($tables as $tmp) {
		if ($tmp->getName() == 'table') {
			$no_iterations ++;
		}
	}
	print "$no_iterations html tables to be processed.\n";
	$interval = intval($no_iterations * 0.01);
	$idx_iterations = 0;

	// Processed the tables
	foreach ($tables as $table) {
        //if (preg_match('/No hits reported/', $table->asXML()) ) {
            //print "skipping this table b/c no hits are reported\n";
        //}
        // make sure we are looking at a table and its not an empty table
        if ($table->getName() == 'table' && !preg_match('/No hits reported/', $table->asXML()) ) {
            $idx_iterations ++;
            if ($idx_iterations % $interval == 0) {
                    $percentage = (int) ($idx_iterations / $no_iterations * 100);
                    tripal_db_set_active($previous_db);
                    tripal_job_set_progress($job_id, $percentage);
                    $previous_db = tripal_db_set_active('chado');
                    print $percentage."% ";
            }

            // Set job status
            // Get the first row and match its name with the feature name
            $firsttd = $table->children()->children()->children();
            $feature_id = 0;
            foreach($firsttd as $b) {
                foreach($b->children() as $a) {
                    if ($a->getName() == 'a') {
                        // Remove _ORF from the sequence name
                        $seqname = preg_replace('/^(.+?)_\d_.+/', "$1", $a);
                        print "seqname is $seqname\n";
    
                        // Find out how many features match this uniquename
                        $sql = "SELECT count(feature_id) FROM {feature} ".
                        "WHERE uniquename = '%s' ";
                        $no_features = db_result(db_query($sql, $seqname));
						    
                        // If there is only one match, get the feature_id
                        if ($no_features == 1) {
                            $sql = "SELECT feature_id FROM {feature} ".
                                    "WHERE uniquename = '%s' ";
                            $feature_id = db_result(db_query($sql, $seqname));
                            print "\tfeature id is $feature_id\n";
    
                            // If the uniquename matches more than one features then skip and print 'Ambiguous'
                        } else if ($no_features > 1) {
                            fwrite($log, "Ambiguous: ".$seqname." matches more than one feature and is not processed.\n");
                            continue;
    
                        // If the uniquename did not match, skip and print 'Failed'
                        } else {
                            fwrite($log, "Failed: ".$seqname."\n");
                        }
    
                    }
                }
            }

            // Successfully matched. print 'Succeeded'. Add analysis_id and
            // feature_id to analysisfeature. Add the table as XML to analysisfeatureprop
            if ($feature_id) {

                //------------------------------------
                // Clease unwanted rows from the table
                //------------------------------------

                $parent_row =   "/<tr><td valign=\"top\"><b>Parent<\/b><\/td>\s*<td valign=\"top\">\s*no.*?parent<\/td>\s*<\/tr>/";
                $children_row = "/<tr><td valign=\"top\"><b>Children<\/b><\/td>\s*<td valign=\"top\">\s*no.*?children<\/td>\s*<\/tr>/";
                $found_row    = "/<tr><td valign=\"top\"><b>Found.*?in<\/b><\/td>\s*<td valign=\"top\">\s*no.*?entries<\/td>\s*<\/tr>/";
                $contains_row = "/<tr><td valign=\"top\"><b>Contains<\/b><\/td>\s*<td valign=\"top\">\s*no.*?entries<\/td>\s*<\/tr>/";
                $go_row       = "/<tr><td valign=\"top\"><b>GO.*?terms<\/b><\/td>\s*<td valign=\"top\">\s*none<\/td>\s*<\/tr>/";
    
                $table_txt = $table->asXML();
                $table_txt = preg_replace($parent_row, "", $table_txt);
                $table_txt = preg_replace($children_row, "", $table_txt);
                $table_txt = preg_replace($found_row, "", $table_txt);
                $table_txt = preg_replace($contains_row, "", $table_txt);
                $table_txt = preg_replace($go_row, "", $table_txt);

                //------------------------------------
                // Clease unwanted ORF link from table 
                //------------------------------------
                $orf_link = "/<b><a href=\"\/iprscan\/wget.*?\">(.*?)<\/a><\/b>/";
                $table_txt = preg_replace($orf_link, "$1", $table_txt);

                //print "----------------------------\n";
                //print "old: ".$table->asXML()."\n\n\n";
                //print "----------------------------\n";
                //print "Fixed: $table_txt\n";
                //print "----------------------------\n";

                //------------------------------------
                // If this feature has already been associated with this analysis, do not reinsert
                // Otherwise, Insert into analysisfeature table
                //------------------------------------
                $sql = "Select analysisfeature_id as id from {analysisfeature} where feature_id = %d and analysis_id = %d";
                $analysisfeature =  db_fetch_object(db_query($sql, $feature_id, $analysis_id));
                if($analysisfeature){ $analysisfeature_id = $analysisfeature->id; }
                if(!$analysisfeature_id){
                    print "inserting analysisfeature\n";
                   $sql = "INSERT INTO {analysisfeature} (feature_id, analysis_id) ".
                        "VALUES (%d, %d)";
                   db_query ($sql, $feature_id, $analysis_id);
                   $sql = "Select analysisfeature_id from {analysisfeature} where feature_id = %d and analysis_id = %d";
                   $analysisfeature =  db_fetch_object(db_query($sql, $feature_id, $analysis_id));
                   $analysisfeature_id = $analysisfeature->id;
                }

                print "analysisfeature_id is $analysisfeature_id (analysis_id = $analysis_id; feature_id = $feature_id)\n";


                // Get the higest rank for this feature_id in analysisfeatureprop table.
                // If the value of the inserting content is not duplicate, add it to  
                // analysisfeaturepro with 'higest_rank + 1'
                $sql = "SELECT MAX(rank) FROM {analysisfeatureprop} AFP ".
                        "INNER JOIN analysisfeature AF ON AF.analysisfeature_id = AFP.analysisfeature_id ".       
                        "WHERE feature_id=%d ".
                        "AND analysis_id=%d ".
                        "AND type_id=%d ";

                $afp =  db_fetch_object(db_query($sql, $feature_id, $analysis_id, $type_id));
                $hi_rank = 0;
                if ($afp) {
                    $hi_rank = $afp->max + 1;
                }
            
				//------------------------------------------------------------
				// Insert interpro html tags into analysisfeatureprop table
				//------------------------------------------------------------
				// Before inserting, make sure it's not a duplicate
				$sql = "SELECT value FROM {analysisfeatureprop} WHERE analysisfeature_id = %d AND type_id = %d";
				$result = db_query($sql, $analysisfeature_id, $type_id);
				$duplicate = 0;
				while ($afp_value = db_fetch_object($result)) {
				    if ($table_txt == $afp_value->value) {
                        $duplicate = 1;
				    }
				}
				if (!$duplicate) {
				    $sql = "INSERT INTO {analysisfeatureprop} (analysisfeature_id, type_id, value, rank)".
                            "VALUES (%d, %d, '%s', %d)";
				    db_query($sql, $analysisfeature_id, $type_id, $table_txt, $hi_rank);
				    fwrite($log, " (Insert)\n"); // write to log
                    print "\twriting table\n";
				} else {
				   fwrite($log, " (Skipped)\n");
                    print "\tskipping table - dup\n";
				}
				
                // Parse GO terms. Make sure GO database schema is installed in chado
                $go_db_id = db_result(db_query("SELECT db_id FROM {db} WHERE name='GO'"));
                if (!$go_db_id) {
                    print 'GO schema not installed in chado. GO terms are not processed.';
                }
                if ($go_db_id && $parsego) {
                    $trs = $table->children();
                    foreach ($trs as $tr) {
                        $tds = $tr->children();
                        foreach($tds as $td) {
                            $gotags = $td->children();
                            foreach ($gotags as $gotag) {
                                // Look for 'GO:accession#'
                                if (preg_match("/^.*?GO:(\d+).*$/", $gotag, $matches)) {
                        	
                                    // Find cvterm_id for the matched GO term
                                    $sql = "SELECT cvterm_id FROM {cvterm} CVT 
                                            INNER JOIN dbxref DBX ON CVT.dbxref_id = DBX.dbxref_id
                                            WHERE DBX.accession = '%s' AND DBX.db_id = %d";
                                    $goterm_id = db_result(db_query($sql, $matches[1], $go_db_id));
                           
                                    //-------------------------------------------
                                    // Insert GO terms into feature_cvterm table
                                    //-------------------------------------------
                                    // Default pub_id = 1 (NULL) was used
                                    $sql = "INSERT INTO {feature_cvterm} (feature_id, cvterm_id, pub_id)
                                            VALUES (%d, %d, 1)";
                                    db_query($sql, $feature_id, $goterm_id);

                                    //------------------------------------------------
                                    // Insert GO terms into analysisfeatureprop table
                                    //------------------------------------------------
                                    $sql = "INSERT INTO {analysisfeatureprop} (analysisfeature_id, type_id, value, rank) ".
                                            "VALUES (%d, %d, '%s', 0)";
                                    db_query($sql, $analysisfeature_id, $goterm_id, $matches[1]);
                                }
                            }
                        }
                    }
                }
            }
        }
    }
    tripal_db_set_active ($previous_db); // Use drupal database
    print "Done.\nSuccessful and failed entries have been saved in the log file:\n $logfile\n";
	
    fwrite($log, "\n");
    fclose($log);
    return;
}

/*******************************************************************************
 * 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_analysis_interpro_form'] = array(
                  '#value' => theme('tripal_analysis_interpro_results', $node),
                  '#weight' => 7
					);
				}
			}
	}
}
/************************************************************************
 *  We need to let drupal know about our theme functions and their arguments.
 *  We create theme functions to allow users of the module to customize the
 *  look and feel of the output generated in this module
 */
function tripal_analysis_interpro_theme () {
	return array(
      'parse_EBI_Interpro_XML_index_version' => array (
         'arguments' => array('node'),
	),
      'tripal_analysis_interpro_results' => array (
         'arguments' => array('node'),
	)
	);
}
/*******************************************************************************
 * Prepare interpro result for the feature shown on the page
 */
function theme_tripal_analysis_interpro_results ($node) {
	$feature = $node->feature;
	$content = tripal_get_interpro_results($feature->feature_id);
	return $content;
}

/*******************************************************************************
 * 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;
}

/*******************************************************************************
 * tripal_get_interpro_results()
 * Get interpro result from featureprop table for the feature
 */
function tripal_get_interpro_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 .= "<div id=\"interpro-hits\" class=\"tripal_interpro-info-box\">
                     <div class=\"tripal_expandableBox\">
                       <h3>InterProScan Analysis</h3>
                     </div>
                     <div class=\"tripal_expandableBoxContent\">
                       <span>
                         <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></span></div></div>';
   }
     
   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;
}