lims.module

<?php
/**
 * @file
 * The bones of the LIMS system.
 */
 require_once 'includes/LIMSrawseq.inc';

/**
 * Implements hook_menu().
 */
function lims_menu() {
  $items['node/%node/download-index'] = array(
    'title' => 'Index File Download',
    'description' => 'For hosting a plain text, tab-delimited mapping of samples to indices.',
    // permissions required to view page
    'access arguments' => array('access content'),
    'page callback' => 'lims_download_samples_indices_callback',
    'page arguments' => array(1),
  );

  return $items;
}

/**
 * Implements hook_permission().
 */
function lims_permission() {
  return array(
    'access lims' => array(
      'title' => t('Access LIMS'),
      'description' => t('Apply access permission to LIMS view page.'),
    )
  );
}

/**
 * Implements hook_node_access().
 */
function lims_node_access($node, $op, $account) {
  if (isset($node->type) AND ($node->type == 'seq_run')) {
    return (user_access('access lims')) ? NODE_ACCESS_ALLOW : NODE_ACCESS_DENY;
  }

  return NODE_ACCESS_IGNORE;
}

/**
 * Page callback function for hook_menu()
 */
function lims_download_samples_indices_callback($node) {
  $output = '';
  if (isset($node->samples[0]->index_seq)) {
    // Remove Knowpulse header and menu
    drupal_add_http_header('Content-Type', 'text/plain');
    // Automatically download as a file
    header("Content-Disposition: attachment; filename=indices.txt");

    // Iterate through each sample for this node and extract the sample name and index sequence
    foreach ($node->samples as $sample) {
      $output .= "$sample->sample_name\t$sample->index_seq\n";
    }
    print $output;

    // This effectively "breaks drupal" (as bad as killing kittens?) right after we get what
    // we need to prevent any theming from appearing on the page
    drupal_exit();
  }
  else {
    // CSS to format the link on the page
    $output .= '<style>
                a.indices-link-download-redirect {
                float:right;
                margin-top: 17px;
                margin-right: 17px;
                } </style>';
    // The error message formated as a drupal error
    $output .= '<div class="messages error">There are no indices for this sequencing run, therefore no file can be downloaded.</div>';
    // Provide a link to return the user to the Sequencing Run for this node
    $output .= l('Return to Sequencing Run', 'node/' . $node->nid, array('attributes' => array('class' => array('indices-link-download-redirect'))));
  }

  return $output;
}

/**
 * Implements hook_theme().
 */
function lims_theme($existing, $type, $theme, $path) {
  $items = array();

  $items['lims_seqrun_node'] = array(
    'template' => 'lims-seqrun-node',
    'render element' => 'node',
    'path' => "$path/theme"
  );

  $items['lims_seqrun_teaser'] = array(
    'template' => 'lims-seqrun-teaser',
    'render element' => 'node',
    'path' => "$path/theme"
  );

  return $items;
}

/**
 * @section
 * "Sequencing Run" Node/Content type.
 */

/**
 * Implements hook_node_info().
 * Register the "Sequencing Run" node/content type with Drupal.
 */
function lims_node_info() {
  $node_types = array();

  $node_types['seq_run'] = array(
    'name' => t('Sequencing Run'),
    'base' => 'seqrun',
    'description' => t('Information about a run of sequencing. This is used to keep track of raw sequence that will not be loading into KnowPulse directly but for which meta data including file locations need to be stored.'),
    'title_label' => t('Run Name'),
    'locked' => TRUE
  );

  return $node_types;
}

/**
 * Implements hook_load().
 * Loads our custom values into the node object for viewing and updating.
 */
function seqrun_load(&$nodes) {

  // Keep track of the sample file ids while populating the main data
  // to make for more efficient retrieval of samples later.
  $sample_fids = array();
  $sample_ids = array();

  // Select all the details to populate the nodes.
  $nids = array_keys($nodes);
  if (count($nids) > 1) {
    $results = db_query('
      SELECT *
      FROM {lims_seqrun} run
      LEFT JOIN {lims_seqrun_species} org ON org.run_id=run.run_id
      WHERE nid IN (:nids)',
      array(':nids' => $nids)
    );
  }
  else {
    $results = db_query('
      SELECT *
      FROM {lims_seqrun} run
      LEFT JOIN {lims_seqrun_species} org ON org.run_id=run.run_id
      WHERE nid=:nid',
      array(':nid' => $nids)
    );
  }

  // Now add them all to the nodes.
  foreach($results as $r) {
    $nodes[$r->nid]->seqrun = (isset($nodes[$r->nid]->seqrun)) ? $nodes[$r->nid]->seqrun : new stdClass();
    $nodes[$r->nid]->seqrun->run_id = $r->run_id;
    $nodes[$r->nid]->seqrun->run_name = $r->run_name;
    $nodes[$r->nid]->seqrun->filename = $r->filename;
    $nodes[$r->nid]->seqrun->md5sum = $r->md5sum;
    $nodes[$r->nid]->seqrun->library_type = $r->library_type;
    $nodes[$r->nid]->seqrun->insert_length = $r->insert_length;
    $nodes[$r->nid]->seqrun->fragment_length = $r->fragment_length;
    $nodes[$r->nid]->seqrun->index_type = $r->index_type;
    $nodes[$r->nid]->seqrun->technology = $r->technology;
    $nodes[$r->nid]->seqrun->read_type = $r->read_type;
    $nodes[$r->nid]->seqrun->description = $r->description;
    $nodes[$r->nid]->seqrun->samples_fid = $r->samples_fid;
    $nodes[$r->nid]->samples = array();

    // Keep track of the sample file id for each loaded node
    // so that we can attach the samples later.
    if ($r->samples_fid) {
      $sample_fids[$r->samples_fid] = $r->nid;
    }

    // Keep track of nodes with single samples
    if ($r->sample_id) {
      $sample_ids[$r->sample_id] = $r->nid;
    }

    // Since there can be more than one species, the join above will actually
    // return multiple rows (r's) for a single node. Thus we need to check if
    // this node already has a species set and if it does then add to not
    // override the array. Above we just override the values since they
    // will not change between rows and that's faster then checking first.
    if (isset($nodes[$r->nid]->species)) {
      $nodes[$r->nid]->species[$r->organism_id] = chado_generate_var('organism', array('organism_id' => $r->organism_id));
    }
    else {
      $nodes[$r->nid]->species = array($r->organism_id => chado_generate_var('organism', array('organism_id' => $r->organism_id)));
    }
  }

  // Now retrieve and add samples.
  // CASE 1: Single Sample to retrieve
  if (sizeof($sample_ids) > 0) {
    $results = db_query('
      SELECT *
      FROM {lims_seqrun_samples}
      WHERE sample_id IN (:ids)
    ', array(':ids' => array_keys($sample_ids)));
    // Add each sample to the appropriate node based on the sample fids array
    // we created above which maps fid to nid.
    foreach($results as $r) {
      $nid = $sample_ids[$r->sample_id];

      $sample = array(
        'sample_id' => $r->sample_id,
        'sample_name' => $r->sample_name,
        'sample_accession' => $r->sample_accession,
        'sample_stock_id' => $r->sample_stock_id,
        'sample_description' => $r->sample_description,
      );
      $nodes[$nid]->samples[] = (object) $sample;
    }
  }
  // CASE 2: Multiple samples grouped by a sample_fid.
  if (sizeof($sample_fids) > 0) {
    $results = db_query('
      SELECT *
      FROM {lims_seqrun_samples}
      WHERE samples_fid IN (:fids)
      ORDER BY sample_id ASC
    ', array(':fids' => array_keys($sample_fids)));
    // Add each sample to the appropriate node based on the sample fids array
    // we created above which maps fid to nid.
    foreach($results as $r) {
      $nid = $sample_fids[$r->samples_fid];
      $sample = array(
        'sample_id' => $r->sample_id,
        'index_id' => $r->index_id,
        'index_seq' => $r->index_seq,
        'sample_name' => $r->sample_name,
        'sample_accession' => $r->sample_accession,
        'sample_stock_id' => $r->sample_stock_id,
        'sample_description' => $r->sample_description,
        'plate_well' => $r->plate_well,
      );
      $nodes[$nid]->samples[] = (object) $sample;
    }
  }

  // Add quality Information to any samples which were collected.
  foreach ($nodes as $nid => $node) {
    foreach ($node->samples as $k => $sample) {
      $sample_id = $sample->sample_id;
      $results = db_select('lims_seqrun_samples_info', 'info')
        ->fields('info', ['sample_info_id', 'info_label', 'value'])
        ->condition('info.sample_id', $sample_id)
        ->orderBy('info.sample_info_id', 'ASC')
        ->execute();
      foreach($results as $r) {
        $node->samples[$k]->quality_info[$r->info_label] = $r->value;
      }
    }
  }
}

/**
 * Implement hook_form().
 * Creates the node/page add/edit form for a "Sequencing Run".
 */
function seqrun_form($node, $form_state) {
  $form = node_content_form($node, $form_state);

  $form['general'] = array(
    '#type' => 'fieldset'
  );

  $form['title']['#description'] = t('Choose a short, descriptive name for the run.');
  $form['general']['title'] = $form['title'];
  unset($form['title']);

  // Hard-coded options.
  // Change the arrays below to change the options available for each select.
  // NOTE: species are based on those in the chado organism table so cannot
  // be changed manually.
  $options = array(
    'species' => array(),
    'technology' => array(
      'Sanger',
      'Illumina GAIIx',
      'Illumina HiSeq 2500',
      'Illumina MiSeq',
      'Illumina HiSeq X Ten',
      'Illumina NovaSeq6000',
      'Roche 454',
      'PacBio RS',
      'PacBio HiFi',
      'Pacbio Revio',
      'Oxford Nanopore',
    ),
    'read_type' => array(
      'Single-end',
      'Paired-end',
      'Mate Pair (FR)',
      'Mate Pair (RF)',
      'Other',
    ),
    'library_type' => array(
      'Genomic',
      'RNA-seq',
      '3\' transcript',
      'Single-enzyme GBS',
      'Dual-enzyme GBS',
      'ChIPseq',
      'Exome capture',
      '10X Chromium',
      'Oxford Nanopore: ligation',
      'Oxford Nanopore: rapid',
      'Oxford Nanopore: PCR-cDNA',
      'CRISPRclean Ribosomal RNA Depletion (Jumpcode Genomics)',
    ),
    'index_type' => array(
      'None',
      'Illumina indexing',
      'GBS 96-plex',
      'NextFlex 96-plex',
      '10X Chromium index set',
      'Custom',
      'Other',
    ),
  );
  // Add Species options.
  $results = chado_query("SELECT organism_id, genus||' '||species as scientific_name FROM {organism} ORDER BY genus ASC, species ASC");
  foreach ($results as $r) {
    $options['species'][$r->organism_id] = $r->scientific_name;
  }

  $form['run_id'] = array(
    '#type' => 'hidden',
    '#value' => (isset($node->seqrun->run_id)) ? $node->seqrun->run_id : NULL,
  );

  // Ensure keys = values for options.
  foreach ($options as $category => $opts) {
    if ($category != 'species') {
      $options[$category] = array_merge(
        // Add NULL option to force users to select.
        array(NULL => ''),
        // Use the previous values as both the keys and values.
        array_combine($opts, $opts)
      );
    }
  }

  // Add a little bit of theme-ing for the form.
  $form['#attached']['css'][] = array(
    'type' => 'inline',
    'data' => '
      select {
        width: 370px;
      }
      em {
        font-weight: bold;
        font-style: italic;
      }
      .footnotes {
        margin-left: 25px;
      }
    '
  );
  // This is needed to remove error messages associated with a previously
  // uploaded file which has since been removed.
  $form['#attached']['js'][] = array(
    'type' => 'inline',
    'data' => "
      Drupal.behaviors.limsSeqRun_nodeFormSamplesErrorRemove = {
          attach: function (context, settings) {

            if(document.getElementById('edit-sample-file-upload-button')) {
              jQuery('div.messages.error').remove();
            }
      }};
    "
  );

  $form['general']['species'] = array(
    '#type' => 'select',
    '#title' => 'Species',
    '#description' => 'Please select the species the sequencing was done on. If your samples are from multiple species, then ctrl + click to select all species represented. Furethermore, if your species is not in this list, please contact Larissa or Lacey before entering data.',
    '#options' => $options['species'],
    '#multiple' => TRUE,
    '#required' => TRUE,
    '#default_value' => (isset($node->species)) ? array_keys($node->species) : NULL,
  );

  $form['library'] = array(
    '#type' => 'fieldset',
    '#title' => t('Library-Specific Details'),
  );

  $form['library']['library_type'] = array(
    '#type' => 'select',
    '#title' => t('Library Type'),
    '#description' => t('The method of library preparation for sequencing.'),
    '#options' => $options['library_type'],
    '#required' => TRUE,
    '#default_value' => (isset($node->seqrun->library_type)) ? $node->seqrun->library_type : '',
  );

  $form['library']['insert_len'] = array(
    '#type' => 'textfield',
    '#title' => t('Insert Length'),
    '#description' => 'Enter the length of the insert in base pairs. This will not apply to all sequencing technologies.',
    '#element_validate' => array('element_validate_integer_positive'),
    '#default_value' => (!empty($node->seqrun->insert_length)) ? $node->seqrun->insert_length : '',
  );

  $form['library']['frag_len'] = array(
    '#type' => 'textfield',
    '#title' => t('Fragment Length'),
    '#description' => 'Enter the length of the fragment in base pairs. This will not apply to all sequencing technologies.',
    '#element_validate' => array('element_validate_integer_positive'),
    '#default_value' => (!empty($node->seqrun->fragment_length)) ? $node->seqrun->fragment_length : '',
  );


  $form['library']['index_type'] = array(
    '#type' => 'select',
    '#title' => t('Index Type'),
    '#description' => t('The collection of barcodes used to identify samples. If "Other", please elaborate in the "Run Description".'),
    '#options' => $options['index_type'],
    '#required' => TRUE,
    '#default_value' => (isset($node->seqrun->index_type)) ? $node->seqrun->index_type : '',
    '#ajax' => array(
      'callback' => 'ajax_lims_seqrun_index_type_callback',
      'wrapper' => 'samples-section',
    ),
  );

  $form['sequencing'] = array(
    '#type' => 'fieldset',
    '#title' => t('Sequencing-Specific Details')
  );

  $form['sequencing']['technology'] = array(
    '#type' => 'select',
    '#title' => 'Sequencing Technology',
    '#description' => 'Choose the sequencing platform.',
    '#options' => $options['technology'],
    '#required' => TRUE,
    '#default_value' => (isset($node->seqrun->technology)) ? $node->seqrun->technology : '',
  );

  $form['sequencing']['read_type'] = array(
    '#type' => 'select',
    '#title' => t('Read Type'),
    '#description' => t('Choose the method used to generate the reads.'),
    '#options' => $options['read_type'],
    '#required' => TRUE,
    '#default_value' => (isset($node->seqrun->read_type)) ? $node->seqrun->read_type : '',
  );

  $form['file'] = array(
    '#type' => 'fieldset',
    '#title' => 'File-Specific Details',
  );

  $form['file']['filename'] = array(
    '#type' => 'textfield',
    '#title' => 'File Name',
    '#description' => 'Specify the full name of the file <em>including the path and extension</em>.',
    '#default_value' => (isset($node->seqrun->filename)) ? $node->seqrun->filename : '',
  );

  $form['file']['md5sum'] = array(
    '#type' => 'textfield',
    '#title' => 'MD5 Checksum',
    '#description' => 'The MD5 hash describing the file (used to verify the file is complete & intact).',
    '#default_value' => (isset($node->seqrun->md5sum)) ? $node->seqrun->md5sum : '',
  );

  $form['sample'] = array(
    '#type' => 'fieldset',
    '#title' => t('Sample Details'),
    '#prefix' => '<div id="samples-section">',
    '#suffix' => '</div>'
  );

  $index_type = (isset($node->seqrun->index_type)) ? $node->seqrun->index_type : NULL;
  $index_type = (isset($form_state['values']['index_type'])) ? $form_state['values']['index_type'] : $index_type;

  // Samples can be submitted as either a single sample or a file containing multiple samples
  // depending on whether they are indexed or not. Unfortunatly doing a simple if indexed
  // then single sample otherwise, file upload causes Drupal Magic to go awry... Thus we
  // are assuming file upload to appease the Drupal Gods but if there is no indexing we
  // we will later delete the file upload elements in favour of single sample entry fields.

  // File Upload for indexed samples.
  // --------------------------------------------
  $warning = '<div class="messages warning">
      <h2 class="element-invisible">Warning message</h2>
      File Upload is specific to multiplexed samples. If you only have a single sample, select "None" as the "Index Type" above.
    </div>';
  $xls_template = drupal_get_path('module','lims') . '/LIMS-Sequence_Run-samples_template.xls';
  $tsv_template = drupal_get_path('module','lims') . '/LIMS-Sequence_Run-samples_template.tsv';
  $help = '<br /><strong>Samples File Template:</strong> <a href="' . url($xls_template) . '">XLS</a>, <a href="' . url($tsv_template) . '">TSV</a><br /><br />
    <strong>Samples File Specifications:</strong><br />
    A tab-delimited file where each line describes a single well/lane of sequencing. Each line should have the the following columns:
    <ol>
      <li><em>Index ID<span style="color:red" title="required">*</span>:</em> The name of the index used for this particular sample (ie: GBS-Idx1).</li>
      <li><em>Index Sequence<span style="color:red" title="required">*</span>:</em> The sequence of the index (ie: CTCC).</li>
      <li><em>Sample Name<span style="color:red" title="required">*</span>:</em> The name you gave your sample. NOTE: IF your sample name exactly matches that of the germplasm you extracted it from then the KP:GERM ID will be looked-up automatically.</li>
      <li><em>Sample Accession<span style="color:red" title="required"><sup>++</sup></span>:</em> The KP:GERM ID in KnowPulse for the germplasm your sample was extracted from (ie: KP:GERM58 if your sample is CDC Redberry).</li>
      <li><em>Plate Well:</em> The location of the sample in the plate (ie: A1).</li>
      <li><em>Sample Description:</em> Any additional notes or description of the sample (ie: seed source, number of plants pooled, etc.).</li>
    </ol>
      <div class="footnotes"><span style="color:red" title="required">* Required Columns.</span></div>
      <div class="footnotes"><span style="color:red" title="required"><sup>++</sup> Accession is required unless your sample name exactly matches germplasm in KnowPulse. The accession in the file (if supplied) will always take precedence.</span></div>
      <div class="footnotes"><em>NOTE: The file should contain 6 columns. If the column is not required then leave it empty.</em></div>
    <br />';
  $form['sample']['file_help'] = array(
    '#type' => 'markup',
    '#markup' => $warning.$help
  );

  // Needed for file upload.
  $form['#attributes']['enctype'] = 'multipart/form-data';

  // Provide field to upload file.
  $samples_fid = (isset($node->seqrun->samples_fid)) ? $node->seqrun->samples_fid : NULL;
  $samples_fid = (isset($form_state['values']['sample_file'])) ? $form_state['values']['sample_file'] : $samples_fid;
  $samples_fid = (isset($form_state['values']) AND empty($form_state['values']['sample_file'])) ? 0 : $samples_fid;
  $form['sample']['sample_file'] = array(
    '#type' => 'managed_file',
    '#title' => 'Samples File',
    '#process' => array('seqrun_file_element_process'),
    '#upload_validators' => array('file_validate_extensions' => array('txt tsv')),
    '#default_value' => $samples_fid,
    '#ajax' => array(
      'callback' => 'ajax_samples_file_show_snippet_callback',
      'wrapper' => 'samples-table',
    )
  );

  $form['sample']['sample_snippet'] = array(
    '#type' => 'markup',
    '#markup' => ''
  );

  if (!empty($samples_fid)) {
    // Generate Samples Snippet to allow users to evaluate
    // whether their file was parsed correctly.
    $header = array(
      array('data' => 'Index', 'colspan' => 2, 'style' => 'text-align:center;'),
      array('data' => 'Sample', 'colspan' => 4, 'style' => 'text-align:center;'),
    );
    $rows = array();
    // Second header.
    $rows[] = array(
      array( 'data' => 'ID', 'header' => TRUE),
      array( 'data' => 'Sequence', 'header' => TRUE),
      array( 'data' => 'Name', 'header' => TRUE),
      array( 'data' => 'Accession', 'header' => TRUE),
      array( 'data' => 'Plate Well', 'header' => TRUE),
      array( 'data' => 'Description', 'header' => TRUE),
    );
    $i = 0;
    $num_samples = db_query('SELECT count(*) FROM {lims_seqrun_samples} WHERE samples_fid=:fid', array(':fid' => $samples_fid))->fetchField();
    $sql = 'SELECT * FROM {lims_seqrun_samples} WHERE samples_fid=:fid ORDER BY sample_id ASC';
    $samples = db_query($sql, array(':fid' => $samples_fid));
    foreach ($samples as $sample) {
      $i++;
      if ($i >= 20) { break; }
      $rows[] = array(
        $sample->index_id,
        $sample->index_seq,
        $sample->sample_name,
        $sample->sample_accession,
        $sample->plate_well,
        $sample->sample_description
      );
    }
    if ($num_samples > 0) {
      $form['sample']['sample_snippet']['#markup'] =
          '<br /><br /><strong>Table 1: A subset of parsed samples from the uploaded file.</strong>'
          . '<br />Please use the following truncated sample list to ensure your file has been parsed correctly.'
          . theme('table',array(
            'header' => $header,
            'rows' => $rows,
            'attributes' => array('style' => 'width:auto')
          ))
          .'<strong>Total Samples Parsed: ' . $num_samples . '</strong> (only a subset of which are shown above)';
    }
  }

  // Input fields for Single Samples.
  // --------------------------------------------
  // If the samples are un-indexed then we can use simple form fields instead of the
  // complicated file upload we just implemented. As such we are going to remove said
  // complicated file upload and replace with more user-friendly fields.
  // If you are confused as to why the file upload needed to be set if it wasn't going to
  // be used, ask the Drupal Gods or see the node above the file upload section if they
  // are not forthcomming.
  if ($index_type == 'None') {

    unset($form['sample']['sample_file']);
    $form['sample']['file_help']['#type'] = 'hidden';
    $form['sample']['sample_snippet']['#type'] = 'hidden';

    // Retrieve the default.
    if (!empty($node->samples) AND !isset($node->samples[0]->index_id)) {
      $sample = $node->samples[0];
      $sample->germplasm_name = str_replace(array(':id',':name'), array($sample->sample_stock_id, $sample->sample_accession), '(:id) :name');
    }
    elseif (isset($form_state['values']) AND isset($form_state['values']['sample_name'])) {
      $sample = new stdclass();
      $sample->sample_name = $form_state['values']['sample_name'];
      $sample->germplasm_name = $form_state['values']['germplasm_name'];
      $sample->sample_description = $form_state['values']['sample_description'];
    }
    else {
      $sample = new stdclass();
      $sample->sample_name = '';
      $sample->germplasm_name = '';
      $sample->sample_description = '';
    }

    $form['sample']['sample_name'] = array(
      '#type' => 'textfield',
      '#title' => 'Sample Name',
      '#description' => 'The name you gave your sample.',
      '#default_value' => $sample->sample_name,
    );

    $form['sample']['germplasm_name'] = array(
      '#type' => 'textfield',
      '#title' => 'Germplasm Name',
      '#description' => 'The name of the germplasm your sample was extracted from.',
      '#autocomplete_path' => 'tripal_ajax/tripal_germplasm/name_to_id/ALL',
      '#default_value' => $sample->germplasm_name,
    );

    $form['sample']['sample_description'] = array(
      '#type' => 'textarea',
      '#rows' => 2,
      '#title' => 'Sample Description',
      '#description' => 'Any additional notes or description of the sample (ie: seed source, number of plants pooled, etc.).',
      '#default_value' => $sample->sample_description,
    );

  }

  $form['description'] = array(
    '#type' => 'textarea',
    '#title' => t('Run Description'),
    '#description' => t('Any additional details related to this sequencing run.'),
    '#default_value' => (isset($node->seqrun->description)) ? $node->seqrun->description : '',
  );

  return $form;
}

/**
 * AJAX: Replace samples section based on index type.
 */
function ajax_lims_seqrun_index_type_callback($form, $form_state) {
  return $form['sample'];
}

/**
 * A custom process function for the seqrun sample details file form element.
 */
function seqrun_file_element_process($element, &$form_state, &$form) {

  // Pre-render the managed file field in order to gain access to the button.
  $element = file_managed_file_process($element, $form_state, $form);

  // Move the samples table into the file element so it can piggy-back
  // on the file upload/remove ajax.
  $element['sample_snippet'] = $form['sample']['sample_snippet'];
  unset($form['sample']['sample_snippet']);

  // Add a validate handler so we can do stuff!
  $element['upload_button']['#submit'][] = 'seqrun_sample_file_submit';
  $element['remove_button']['#submit'][] = 'seqrun_sample_file_remove_submit';

  return $element;
}

/**
 * Custom handler for Sequencing Run samples file.
 */
function seqrun_sample_file_submit($form, &$form_state) {
  $error = FALSE;

  ini_set('auto_detect_line_endings', true);

  // First save the file.
  $file = file_save_upload('sample_file');

  // As long as the file was saved properly (ie: no validation issues so far)
  // then we can start processing it.
  if (!empty($file)) {

    // Open the file for processing.
    $sample_file = fopen($file->uri, 'r');
    $header = fgetcsv($sample_file, 0, "\t");
    if (sizeof($header) < 6) {
      $error = TRUE;
      drupal_set_message('Six columns are required for every line.', 'error');
      //form_set_error('sample_file', 'Line '.$i.': 6 columns are required for every line.');
    }

    // For each line:
    $i = 1;
    while( $line = fgetcsv($sample_file, 0, "\t") ) {
      $i++;
      $error = FALSE;
      if (is_array($line)) {

        // Reset extra fields
        $sample_stock_id = NULL;

        // Trim whitespace for each column.
        foreach ($line as $k => $v) {
          $line[$k] = trim($v);
        }

        // Determine if the line is empty.
        $empty = FALSE;
        if (sizeof($line) == 1 AND empty($line[0])) {
          $empty = TRUE;
        }
        // Validate the line.
        if (!$empty) {
          // -- Check no column is longer than the database allows.
          $expected = array(80, 80, 80, 80, 10);
          foreach ($expected as $k => $max) {
            if (isset($line[$k]) AND (strlen($line[$k]) > $max)) {
              $error = TRUE;
              $column = ($k == 0) ? 'Index ID' : NULL;
              $column = ($k == 1) ? 'Index Sequence' : $column;
              $column = ($k == 2) ? 'Sample Name' : $column;
              $column = ($k == 3) ? 'Sample Accession' : $column;
              $column = ($k == 4) ? 'Plate Well' : $column;
              drupal_set_message('Line '.$i.':  The <i>'.$column.'</i> is too long (maximum length: '.$max.' characters).', 'error');
            }
          }
          // -- Index ID not empty.
          if (empty($line[0])) {
            $error = TRUE;
            drupal_set_message('Line '.$i.': The Index ID is required and thus should not be empty.','error');
            //form_set_error('sample_file', 'Line '.$i.': The first three columns (Index ID, Index Sequence & Sample Name) are required and thus should not be empty.');
          }
          // -- Index Sequence not empty
          //    EXCEPTION: Index ID is "unknown".
          if (empty($line[1]) AND !preg_match('/[uU]nknown/',$line[0])) {
            $error = TRUE;
            drupal_set_message('Line '.$i.': The Index Sequence required and thus should not be empty.','error');
            //form_set_error('sample_file', 'Line '.$i.': The first three columns (Index ID, Index Sequence & Sample Name) are required and thus should not be empty.');
          }
          // -- Sample Name not empty.
          if (empty($line[2])) {
            $error = TRUE;
            drupal_set_message('Line '.$i.': The Sample Name is required and thus should not be empty.','error');
            //form_set_error('sample_file', 'Line '.$i.': The first three columns (Index ID, Index Sequence & Sample Name) are required and thus should not be empty.');
          }
          // -- 2nd column is all sequence.
          //    EXCEPTION: Index ID is "unknown".
          if (!empty($line[1]) AND !preg_match('/^[ATGCatgc\,]+$/', $line[1]) AND !preg_match('/[uU]nknown/',$line[0])) {
            $error = TRUE;
            drupal_set_message('Line '.$i.': The index sequence should consist of only ATGC.','error');
            //form_set_error('sample_file', 'Line '.$i.': The index sequence should consist of only ATGC.');
          }
          // -- Accession should always be KP:GERM###.
          if (!empty($line[3]) AND !preg_match('/KP:GERM\d+/', $line[3])) {
            $error = TRUE;
            drupal_set_message('Line '.$i.': The accession should always be KP:GERM### in full capitals.','error');
          }
          // -- Ensure KP:GERM exists.
          elseif (!empty($line[3])) {
            $kpgerm = chado_query('SELECT stock_id FROM {stock} WHERE uniquename=:accession', array(':accession' => $line[3]))->fetchField();
            if (empty($kpgerm)) {
              $error = TRUE;
              drupal_set_message('Line '.$i.': The accession provided, <i>'.$line[3].'</i>, does not exist in KnowPulse.', 'error');
            }
            else {
              $sample_stock_id = $kpgerm;
            }
          }
          // -- Attempt Auto-detect Accession based on sample name.
          else {

            // First TRY to look it up based on sample name.
            $kpgerm = chado_query('SELECT uniquename, stock_id FROM {stock} WHERE name=:sample', array(':sample' => $line[2]))->fetchObject();
            if (!empty($kpgerm)) {
              $line[3] = $kpgerm->uniquename;
              $sample_stock_id = $kpgerm->stock_id;
            }
            // If unable to then provide error message. :(
            else {
              $error = TRUE;
              drupal_set_message('Line '.$i.': The accession is required and could not be automatically determined by the sample name. Please provide it in the file.','error');
              //form_set_error('sample_file', 'Line '.$i.': The accession should always be KP:GERM### in full capitals.');
            }
          }

          // Save it to the database.
          if (!$error) {
            $sample = array(
              'samples_fid' => $file->fid,
              'index_id' => $line[0],
              'index_seq' => $line[1],
              'sample_name' => $line[2],
              'sample_accession' => $line[3],
              'plate_well' => (isset($line[4])) ? $line[4] : '',
              'sample_description' => (isset($line[5])) ? $line[5] : '',
              'sample_stock_id' => $sample_stock_id,
            );
            drupal_write_record('lims_seqrun_samples', $sample);
          }
        }
      }
    }
  }

}

/**
 * Custom handler for Sequencing Run samples file.
 */
function seqrun_sample_file_remove_submit($form, &$form_state) {

  // Clean-up sample table to ensure that no samples from previously
  // uploaded files hang around.
  db_query('DELETE FROM {lims_seqrun_samples} WHERE samples_fid NOT IN (SELECT samples_fid FROM {lims_seqrun})');

  // Remove the fid from this node.
  if (isset($form_state['node']->nid)) {
    db_query("UPDATE {lims_seqrun} SET samples_fid=NULL WHERE nid=:nid",
      array(':nid' => $form_state['node']->nid));
  }
}

/**
 *
 */
function ajax_samples_file_show_snippet_callback($form, $form_state) {
  return $form['sample']['sample_snippet'];
}

/**
 * Implement hook_validate().
 * Validates the data entered into the create/edit "Sequencing Run" form
 * before either inserting, seqrun_updating(), or updating, seqrun_insert().
 */
function seqrun_validate(&$node, $form, &$form_state) {

  // Ensure the insert & fragment length are integers.
  // Taken care of by #element_validate.

  // Ensure the file name has not already been associated with a sequencing run.
    $filename = NULL;
  if (isset($node->nid) AND !empty($form_state['values']['filename'])) {
    $filename = db_query("SELECT nid from {lims_seqrun} WHERE filename = :filename AND nid != :nid LIMIT 1", array(":filename" => $form_state['values']['filename'], ':nid' => $node->nid))->fetchField();
  }
  elseif (!empty($form_state['values']['filename'])) {
    $filename = db_query("SELECT nid from {lims_seqrun} WHERE filename = :filename LIMIT 1", array(":filename" => $form_state['values']['filename']))->fetchField();
  }
  if (!empty($filename)) {
    form_set_error('filename', t('This file has already been associated with <a href="@url">a sequencing run</a>.', array('@url' => 'node/'.$filename)));
  }

  // Ensure the file exists
  // @todo: unsure how to do this when storage is not yet mounted.

  // The following validation only applies to a single unindexed sample.
  if ($node->index_type == 'None' AND isset($node->germplasm_name)) {
    // Ensure the autocomplete was used for the germplasm name.
    if (!preg_match('/\((\d+)\).*/', $node->germplasm_name, $matches)) {
      form_set_error('germplasm_name', 'You must choose the germplasm name from the autocomplete drop-down.');
    }
    // Ensure the stock_id from the autocomplete exists.
    else {

      $form_state['values']['germplasm_id'] = $matches[1];
      $present = chado_query('SELECT true FROM {stock} WHERE stock_id=:id',
        array(':id' => $form_state['values']['germplasm_id'])) ->fetchField();
      if (!$present) {
        form_set_error('germplasm_name', 'The germplasm selected does not exist. Please re-select the <em>Germplasm Name</em> from the drop-down.');
      }

    }
  }

}

/**
 * Implements hook_insert().
 * Handles insertion of our custom content when a new "Sequencing Run"
 * has been created.
 */
function seqrun_insert($node) {

  // If the index type is set to None then we have to
  // insert the single sample since it won't be added with a file.
  // Do this first so that we can reference it in the core Sequencing Run table.
  if ($node->index_type == 'None') {

    preg_match('/\((\d+)\)/', $node->germplasm_name, $matches);

    // NOTE: We're ignoring the sample_fid, index_id, index_seq & plate_well
    // since they don't apply to a single sample sequencing run.
    $sample = array(
      'sample_name' => $node->sample_name,
      'sample_accession' => 'KP:GERM' . $matches[1],
      'sample_description' => $node->sample_description,
      'sample_stock_id' => $node->germplasm_id,
    );
    drupal_write_record('lims_seqrun_samples', $sample);

  }

  // Now add data for the core sequencing run.
  $record = array(
    'nid' => $node->nid,
    'run_name' => $node->title,
    'filename' => $node->filename,
    'md5sum' => $node->md5sum,
    'library_type' => $node->library_type,
    'insert_length' => $node->insert_len,
    'fragment_length' => $node->frag_len,
    'index_type' => $node->index_type,
    'technology' => $node->technology,
    'read_type' => $node->read_type,
    'description' => $node->description,
  );
  if (isset($node->sample_file)) {
    $record['samples_fid'] = $node->sample_file;
  }
  if (isset($sample['sample_id'])) {
    $record['sample_id'] = $sample['sample_id'];
  }
  drupal_write_record('lims_seqrun', $record);

  // Now associate the organisms/species.
  foreach($node->species as $organism_id) {
    $species = array(
      'run_id' => $record['run_id'],
      'organism_id' => $organism_id
    );
    drupal_write_record('lims_seqrun_species', $species);
  }

  // Clean-up sample table to ensure that no samples from previously
  // uploaded files hang around.
  db_query('DELETE FROM {lims_seqrun_samples} WHERE samples_fid NOT IN (SELECT samples_fid FROM {lims_seqrun})');
}

/**
 * Implements hook_update().
 * Handles updating our custom content when a pre-existing "Sequencing Run"
 * is edited.
 */
function seqrun_update($node) {

  // If the index type is set to None then we have to
  // insert the single sample since it won't be added with a file.
  // Do this first so that we can reference it in the core Sequencing Run table.
  if ($node->index_type == 'None') {

    // NOTE: We're ignoring the sample_fid, index_id, index_seq & plate_well
    // since they don't apply to a single sample sequencing run.
    $sample = array(
      'sample_name' => $node->sample_name,
      'sample_accession' => 'KP:GERM' . $node->germplasm_id,
      'sample_description' => $node->sample_description,
      'sample_stock_id' => $node->germplasm_id,
    );

    // Check to see if a sample already exists for this sequencing run
    // and then add or update as appropriate.
    $sample_id = db_query('SELECT sample_id FROM {lims_seqrun} WHERE nid=:nid',
      array(':nid' => $node->nid))->fetchField();
    if ($sample_id) {
      $sample['sample_id'] = $sample_id;
      drupal_write_record('lims_seqrun_samples', $sample, 'sample_id');
    }
    else {
      drupal_write_record('lims_seqrun_samples', $sample);
    }
  }

  // Now update data for the core sequencing run.
  $record = array(
    'nid' => $node->nid,
    'run_id' => $node->run_id,
    'run_name' => $node->title,
    'filename' => $node->filename,
    'md5sum' => $node->md5sum,
    'library_type' => $node->library_type,
    'insert_length' => $node->insert_len,
    'fragment_length' => $node->frag_len,
    'index_type' => $node->index_type,
    'technology' => $node->technology,
    'read_type' => $node->read_type,
    'description' => $node->description,
  );
  if (isset($node->sample_file)) {
    $record['samples_fid'] = ($node->sample_file === 0 AND isset($node->seqrun->samples_fid)) ? $node->seqrun->samples_fid : $node->sample_file;
    $record['sample_id'] = NULL;

    // Clean-up any single samples originally attached to this run.
    db_query('DELETE FROM {lims_seqrun_samples} WHERE sample_id IN (SELECT sample_id FROM {lims_seqrun} WHERE run_id=:id)', array(':id' => $node->run_id));

  }
  if (isset($sample['sample_id'])) {
    $record['samples_fid'] = NULL;
    $record['sample_id'] = $sample['sample_id'];
  }
  drupal_write_record('lims_seqrun', $record, 'run_id');

  // Then remove organisms so we can insert fresh.
  db_query("DELETE FROM {lims_seqrun_species} WHERE run_id=:run_id", array(':run_id' => $node->run_id));

  // Now associate the organisms/species.
  foreach($node->species as $organism_id) {
    $species = array(
      'run_id' => $record['run_id'],
      'organism_id' => $organism_id
    );
    drupal_write_record('lims_seqrun_species', $species);
  }

  // Clean-up sample table to ensure that no samples from previously
  // uploaded files hang around.
  db_query('DELETE FROM {lims_seqrun_samples} WHERE samples_fid NOT IN (SELECT samples_fid FROM {lims_seqrun})');

}

/**
 * Implements hook_delete().
 * Handles deletetion of our custom content when a "Sequencing Run" is deleted.
 */
function seqrun_delete($node) {
  db_query('DELETE FROM {lims_seqrun} WHERE nid=:nid', array(':nid' => $node->nid));
}

/**
 * Implements hook_view().
 */
function seqrun_view($node, $view_mode, $langcode = NULL) {

  if ($view_mode == 'teaser') {
    $node->content['teaser'] = array(
      '#theme' => 'lims_seqrun_teaser',
      '#node' => $node,
    );
  }
  else {
    $node->content['main'] = array(
      '#theme' => 'lims_seqrun_node',
      '#node' => $node,
    );
  }

  return $node;
}

/**
 * Implement hook_chado_custom_search().
 *
 * This hook simply lists the machine name of the searches so that we can find
 * the info hook. We've done this for performance reasons.
 */
function lims_chado_custom_search() {
  $searches = [];

  $searches['LIMSrawseq'] = 'Raw Sequence Search';

  return $searches;
}