tripal/tripal_cv/charts.php

<?php


/**
 * Generates JSON used for generating a Google chart of count data associated
 * with a controlled vocabulary.  An example would be features assigned to
 * Gene Ontology terms.  
 * To generate a chart, the progammer must first create a materialized view that
 * will generate count data for a given controlled vocabulary.  For example, the Tripal
 * Analysis GO creates a materialized view for counting Gene Ontology assignments
 * to features.  This view is created on install of the module and is named
 * 'go_count_analysis'.
 *
 * Next, an HTML 'div' box must be added to the desired page
 * with a class name of 'tripal_cv_chart', and an id of the following format:
 *
 *  tripal_[module_name]_cv_chart_[unique id]
 *
 * where [module_name] is the name of the tripal module (e.g. tripal_analyisis_go)
 * and [unique id] is some unique identifier that the contolling module 
 * recognizes. This string is the $chart_id variable passed as the first argument
 * to the function.  For example, the Tripal GO Analysis module generates 
 * chart ids of the form:
 * 
 *  tripal_analysis_go_cv_chart_10_2_bp
 * 
 * In this case the module that will manage this chart is identified as 'tripal_analysis_go' and within
 * the [unique id] portion contains the
 * organism_id (e.g. 10), analysis_id (e.g. 2) and chart type (bp = biological process). 
 *
 * Second, the programmer must then define a hook in the controlling module for setting
 * some options used to build the chart.  The hook has the form:  hook_cv_chart($chart_id).
 * This hook should accept the full $chart_id as the single parameter.  For the Tripal
 * Analysis GO module the hook is named:  tripal_analysis_go_cv_chart.  
 *
 * The array returned by this hook must have the following fields:  
 *  - count_mview
 *      the name of the materialized view that contains the count data
 *      this materialized view must have at least two columns, one with the cvterm_id
 *      for each term and a second column containing the counts
 *  - cvterm_id_column
 *      the column name in the materialized view that contains
 *      the cvterm_ids
 *  - count_column
 *      the column name in the materialized view that contains the
 *      counts
 *  - filter
 *      an SQL compatible WHERE clause string (whithout the word 'WHERE')
 *      that can be used for filtering the records in the materialized view.
 *  - title
 *      the title for the chart
 *  - type
 *      the type of chart to create (see Google Charts documenation). Leave
 *      blank for a pie chart.
 *  - size 
 *      the dimensions of the chart in pixels (see Google Charts documenations 
 *      for exact size limitations).
 *
 * Example from the tripal_analysis_go module:
 * @code
 *  function tripal_analysis_go_cv_chart($chart_id){
 *  
 *    // The CV module will create the JSON array necessary for buillding a
 *    // pie chart using jgChart and Google Charts.  We have to pass to it
 *    // a table that contains count information, tell it which column 
 *    // contains the cvterm_id and provide a filter for getting the
 *    // results we want from the table.
 *    $organism_id = preg_replace("/^tripal_analysis_go_cv_chart_(\d+)-(\d+)_(bp|cc|mf)$/","$1",$chart_id);
 *    $analysis_id = preg_replace("/^tripal_analysis_go_cv_chart_(\d+)-(\d+)_(bp|cc|mf)$/","$2",$chart_id);
 *    $type        = preg_replace("/^tripal_analysis_go_cv_chart_(\d+)-(\d+)_(bp|cc|mf)$/","$3",$chart_id);
 *  
 *    $sql = "SELECT * FROM {Analysis} WHERE analysis_id = %d";
 *    $previous_db = tripal_db_set_active('chado');  // use chado database
 *    $analysis = db_fetch_object(db_query($sql,$analysis_id));
 *    tripal_db_set_active($previous_db);  // now use drupal database  
 *   
 *    if(strcmp($type,'mf')==0){
 *       $class = 'molecular_function';
 *       $title = "Number of Molecular Function Terms From $analysis->name Analysis";
 *    }
 *    if(strcmp($type,'cc')==0){
 *       $class = 'cellular_component';
 *       $title = "Number of Cellular Component Terms From $analysis->name Analysis";
 *    }
 *    if(strcmp($type,'bp')==0){
 *       $class = 'biological_process';
 *       $title = "Number of Biological Process Terms From $analysis->name Analysis";
 *    }
 *    $options = array(
 *       count_mview      => 'go_count_analysis',
 *       cvterm_id_column => 'cvterm_id',
 *       count_column     => 'feature_count',
 *       filter           => "
 *          CNT.organism_id = $organism_id AND 
 *          CNT.analysis_id = $analysis_id AND 
 *          CNT.cvterm_id IN ( 
 *            SELECT CVTR.subject_id 
 *            FROM {CVTerm_relationship} CVTR 
 *              INNER JOIN CVTerm CVT on CVTR.object_id = CVT.cvterm_id
 *              INNER JOIN CV on CVT.cv_id = CV.cv_id
 *            WHERE CVT.name = '$class' AND  
 *                   CV.name = '$class'
 *          )
 *       ",
 *       type             => 'p',
 *       size             => '550x175',
 *       title            => $title,
 *    );
 *    return $options;
 *  }
 *
 * @endcode
 *
 * @param $chart_id
 *   The unique identifier for the chart
 *
 * @return
 *   JSON array needed for the js caller
 *
 * With these three components (materialized view, a 'div' box with proper CSS class and ID, and a hook_cv_chart)
 * a chart will be created on the page.  There is no need to call this function directly.
 *
 * @ingroup tripal_cv
 */
function tripal_cv_chart($chart_id){
  // parse out the tripal module name from the chart_id to find out 
  // which Tripal "hook" to call:
  $tripal_mod = preg_replace("/^(tripal_.+?)_cv_chart_(.+)$/","$1",$chart_id);
  $callback = $tripal_mod . "_cv_chart";

  // now call the function in the module responsible for the chart to fill out
  // an options array needed by the tripal_cv_count_chart call below.
  $opt = call_user_func_array($callback,array($chart_id));

  // build the JSON array to return to the javascript caller
  $json_arr = tripal_cv_count_chart($opt[count_mview],$opt[cvterm_id_column],
     $opt[count_column],$opt[filter],$opt[title], $opt[type],$opt[size]);
  $json_arr[] = $chart_id;  // add the chart_id back into the json array

  return drupal_json($json_arr);

}

 /**
  * This function generates an array with fields compatible with Google charts used
  * for generating pie charts of counts associated with terms in
  * a controlled vocabulary.  An example would be counts of features assigned
  * to terms in the Sequence Ontology.
  *
  * @param $cnt_table
  *   The name of the table (most likely a materialized view) that
  *   contains count data for each term.  The table must have at least
  *   two columns, one with the cvterm_id for each term, and a second
  *   column with the count (i.e. features assigned the term).
  * @param $fk_column
  *   This is the name of the column in the $cnt_table that holds the 
  *   cvterm_id for each term.
  * @param $cnt_column
  *   The name of the column in the $cnt_table containing the counts
  * @param $filter
  *   An SQL compatible 'where' clause (without the word 'WHERE') used
  *   to filter the records in the $cnt_table
  * @param $title
  *   The title of the chart to be rendered.
  * @param $type
  *   The type of chart to be rendered. The value used here is the same as
  *   the type names for Google charts. Default is p3 (pie chart).
  * @param $size
  *   The size in pixels of the chart to be rendered. Default is 300x75. The
  *   size of the chart is constrained by Google charts.  See the Google
  *   chart documentation for exact limitations.
  *  
  * @return 
  *   An array that has the settings needed for Google Charts to creat the chart.
  *
  * @ingroup tripal_cv
  */
function tripal_cv_count_chart($cnt_table, $fk_column,
   $cnt_column, $filter = null, $title = '', $type = 'p3', $size='300x75') {

   if(!$type){
      $type = 'p3';
   }

   if(!$size){
     $size = '300x75';
   }

   if(!$filter){
      $filter = '(1=1)'; 
   }

   $isPie = 0;
   if(strcmp($type,'p')==0 or strcmp($type,'p3')==0){
      $isPie = 1;
   }
   $sql = "
      SELECT CVT.name, CVT.cvterm_id, CNT.$cnt_column as num_items
      FROM {$cnt_table} CNT 
       INNER JOIN {cvterm} CVT on CNT.$fk_column = CVT.cvterm_id 
      WHERE $filter
   ";    

   $features = array();
   $previous_db = tripal_db_set_active('chado');  // use chado database
   $results = db_query($sql);
   tripal_db_set_active($previous_db);  // now use drupal database
   $data = array();
   $axis = array();
   $legend = array();
   $total = 0;
   $max = 0;
   $i = 1;
   while($term = db_fetch_object($results)){
      
      if($isPie){
         $axis[] = "$term->name (".number_format($term->num_items).")";
         $data[] = array($term->num_items,0,0);
      } else {
         $axis[] = "$term->name (".number_format($term->num_items).")";
         $data[] = array($term->num_items);
    //     $legend[] = "$term->name (".number_format($term->num_items).")";
      }
      if($term->num_items > $max){
         $max = $term->num_items;
      }
      $total += $term->num_items;
      $i++;
   }
   // convert numerical values into percentages
   foreach($data as &$set){
      $set[0] = ($set[0] / $total) * 100;
   }
   $opt[] = array(
      data => $data,
      axis_labels => $axis, 
      legend => $legend,
      size => $size, 
      type => $type,
 
      bar_width     => 10, 
      bar_spacing   => 0, 
      title         => $title
   );
//   $opt[] = $sql;
   
   return $opt;
}