File

tripal_feature/includes/gff_loader.inc, line 1810

@todo Add file header description

Code

function tripal_feature_load_gff3_fasta($fh, $interval, &$num_read, &$intv_read, &$line_num) {
  print "Loading FASTA sequences\n";
  $residues = '';
  $sql = " 
    PREPARE sel_gfftemp_un (text) AS
    SELECT feature_id FROM tripal_gff_temp
    WHERE uniquename = $1
  ";
  $status = tripal_core_chado_prepare('sel_gfftemp_un', $sql, array('text'));
  if (!$status) {
    watchdog('T_gff3_loader', 'Cannot prepare statement \'sel_gfftemp_un\'.', 
    array(), WATCHDOG_ERROR);
    return '';
  }
  $id = NULL;

  // iterate through the remaining lines of the file
  while ($line = fgets($fh)) {

    $line_num++;
    $size = drupal_strlen($line);
    $num_read += $size;
    $intv_read += $size;

    $line = trim($line);

    // update the job status every 1% features
    if ($job and $intv_read >= $interval) {
      $intv_read = 0;
      $percent = sprintf("%.2f", ($num_read / $filesize) * 100);
      print "Parsing Line $line_num (" . $percent . "%). Memory: " . number_format(memory_get_usage()) . " bytes.\r";
      tripal_job_set_progress($job, intval(($num_read / $filesize) * 100));
    }

    // if we encounter a definition line then get the name, uniquename,
    // accession and relationship subject from the definition line
    if (preg_match('/^>/', $line)) {
      // if we are beginning a new sequence then save the last one we 
      // just finished.     

      if ($id) {
        $sql = "EXECUTE sel_gfftemp_un('%s')";
        $result = tripal_core_chado_execute_prepared('sel_gfftemp_un', $sql, array($id));
        if (!$result) {
          watchdog('T_gff3_loader', 'Cannot find feature to assign FASTA sequence: %uname', 
          array('%uname' => $id), WATCHDOG_WARNING);
        }
        // if we have a feature then add the residues
        else {
          $feature = db_fetch_object($result);
          $values = array('residues' => $residues);
          $match = array('feature_id' => $feature->feature_id);
          $options = array('statement_name' => 'upd_feature_re');
          tripal_core_chado_update('feature', $match, $values, $options);
        }
      }
      // get the feature ID for this ID from the tripal_gff_temp table
      $id = preg_replace('/^>(.*)$/', '\1', $line);
      $residues = '';
    }
    else {
      $residues .= trim($line);
    }
  }
  // add in the last sequence
  $sql = "EXECUTE sel_gfftemp_un('%s')";
  $result = tripal_core_chado_execute_prepared('sel_gfftemp_un', $sql, array($id));
  if (!$result) {
    watchdog('T_gff3_loader', 'Cannot find feature to assign FASTA sequence: %uname', 
    array('%uname' => $id), WATCHDOG_WARNING);
  }
  // if we have a feature then add the residues
  else {
    $feature = db_fetch_object($result);
    $values = array('residues' => $residues);
    $match = array('feature_id' => $feature->feature_id);
    $options = array('statement_name' => 'upd_feature_re');
    tripal_core_chado_update('feature', $match, $values, $options);
  }
}