|
|
@@ -0,0 +1,977 @@
|
|
|
+<?php
|
|
|
+/**
|
|
|
+ * @file
|
|
|
+ * Contains more generally applicable functions as well as some meant to help developers
|
|
|
+ * Plug-in to the BLAST UI functionality
|
|
|
+ */
|
|
|
+
|
|
|
+namespace Drupal\blast_ui\Services;
|
|
|
+
|
|
|
+use Drupal\node\Entity\Node;
|
|
|
+
|
|
|
+/**
|
|
|
+ * Class TripalD3ColorScheme
|
|
|
+ */
|
|
|
+class BlastUIBlast {
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Get a specific BlastDB.
|
|
|
+ *
|
|
|
+ * @param $identifiers
|
|
|
+ * An array of identifiers used to determine which BLAST DB to retrieve.
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * A fully-loaded BLAST DB Node
|
|
|
+ */
|
|
|
+ public function getBLASTDatabase($identifiers) {
|
|
|
+ $node = FALSE;
|
|
|
+
|
|
|
+ if (isset($identifiers['nid'])) {
|
|
|
+ $node = Node::load($identifiers['nid']);
|
|
|
+ }
|
|
|
+ elseif (isset($identifiers['name'])) {
|
|
|
+ $nid = db_query('SELECT nid FROM {blastdb} WHERE name=:name', [':name' => $identifiers['name']])
|
|
|
+ ->fetchField();
|
|
|
+ $node = Node::load($nid);
|
|
|
+
|
|
|
+ } elseif (isset($identifiers['path'])) {
|
|
|
+ $nid = db_query('SELECT nid FROM {blastdb} WHERE path LIKE :path', [':path' => db_like($identifiers['path']) . '%'])
|
|
|
+ ->fetchField();
|
|
|
+ $node = Node::load($nid);
|
|
|
+ }
|
|
|
+
|
|
|
+ return $node;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Returns a list BLAST DATABASE options
|
|
|
+ *
|
|
|
+ * @param $type
|
|
|
+ * The type of BLAST dabases to restrict the list to (ie: n: nucleotide or p: protein)
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * An array where the nid is the key and the value is the human-readable name of the option
|
|
|
+ */
|
|
|
+ public function getBLASTDatabaseOptions($type) {
|
|
|
+ global $user;
|
|
|
+
|
|
|
+ // Get all BlastDB nodes
|
|
|
+ $nodes = getBLASTDatabaseNodes();
|
|
|
+
|
|
|
+ // Support obsolete database type n/p
|
|
|
+ $obs_type = '';
|
|
|
+ if ($type == 'protein') {
|
|
|
+ $obs_type = 'p';
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $obs_type = 'n';
|
|
|
+ }
|
|
|
+
|
|
|
+ $options = [];
|
|
|
+ foreach ($nodes as $node) {
|
|
|
+ if ( isset($node) && isset($node->db_dbtype) ) {
|
|
|
+ if ( ($node->db_dbtype == $type) OR ($node->db_dbtype == $obs_type) ) {
|
|
|
+ $options[$node->nid] = $node->db_name;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // Sort alphabetically
|
|
|
+ asort($options);
|
|
|
+
|
|
|
+ return $options;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Returns all blast database nodes.
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * An array of nodes.
|
|
|
+ */
|
|
|
+ public function getBLASTDatabaseNodes() {
|
|
|
+ // Use the Entity API to get a list of BLAST Nodes to load
|
|
|
+ // We use this function in order respect node access control so that
|
|
|
+ // administrators can use this module in combination with a node access module
|
|
|
+ // of their choice to limit access to specific BLAST databases.
|
|
|
+ $query = \Drupal::entityQuery('node')
|
|
|
+ // Restrict to BLASTDB nodes.
|
|
|
+ ->condition('bundle', 'blastdb')
|
|
|
+ // Restrict to Published nodes.
|
|
|
+ ->condition('status', 1)
|
|
|
+ // Restrict to nodes the current user has permission to view.
|
|
|
+ ->addTag('node_access');
|
|
|
+
|
|
|
+ $entities = $query->execute();
|
|
|
+
|
|
|
+ // Get all BlastDB nodes
|
|
|
+ return Node::loadMultiple(array_keys($entities['node']));
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Retrieve all the information for a blast job in a standardized node-like format.
|
|
|
+ *
|
|
|
+ * @param $job_id
|
|
|
+ * The non-encoded tripal job_id.
|
|
|
+ * @retun
|
|
|
+ * An object describing the blast job.
|
|
|
+ */
|
|
|
+ public function getBLASTJob($job_id) {
|
|
|
+ $blastjob = db_query('SELECT * FROM blastjob WHERE job_id=:id', array(':id' => $job_id))
|
|
|
+ ->fetchObject();
|
|
|
+
|
|
|
+ if (!$blastjob) {
|
|
|
+ return false;
|
|
|
+ }
|
|
|
+
|
|
|
+ $tripal_job = tripal_get_job($job_id);
|
|
|
+
|
|
|
+ $job = new stdClass();
|
|
|
+ $job->job_id = $job_id;
|
|
|
+ $job->program = $blastjob->blast_program;
|
|
|
+ $job->options = unserialize($blastjob->options);
|
|
|
+ $job->date_submitted = $tripal_job->submit_date;
|
|
|
+ $job->date_started = $tripal_job->start_time;
|
|
|
+ $job->date_completed = $tripal_job->end_time;
|
|
|
+
|
|
|
+ // TARGET BLAST DATABASE.
|
|
|
+ // If a provided blast database was used then load details.
|
|
|
+ if ($blastjob->target_blastdb ) {
|
|
|
+ $job->blastdb = $this->getBLASTDatabase(['nid' => $blastjob->target_blastdb]);
|
|
|
+ }
|
|
|
+ // Otherwise the user uploaded their own database so provide what information we can.
|
|
|
+ else {
|
|
|
+ $job->blastdb = new stdClass();
|
|
|
+ $job->blastdb->db_name = 'User Uploaded';
|
|
|
+ $job->blastdb->db_path = $blastjob->target_file;
|
|
|
+ $job->blastdb->linkout = new stdClass();
|
|
|
+ $job->blastdb->linkout->none = TRUE;
|
|
|
+
|
|
|
+ if ($job->program == 'blastp' OR $job->program == 'tblastn') {
|
|
|
+ $job->blastdb->db_dbtype = 'protein';
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $job->blastdb->db_dbtype = 'nucleotide';
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // FILES.
|
|
|
+ $job->files = new stdClass();
|
|
|
+ $job->files->query = $blastjob->query_file;
|
|
|
+ $job->files->target = $blastjob->target_file;
|
|
|
+ $job->files->result = new stdClass();
|
|
|
+ $job->files->result->archive = $blastjob->result_filestub . '.asn';
|
|
|
+ $job->files->result->xml = $blastjob->result_filestub . '.xml';
|
|
|
+ $job->files->result->tsv = $blastjob->result_filestub . '.tsv';
|
|
|
+ $job->files->result->html = $blastjob->result_filestub . '.html';
|
|
|
+ $job->files->result->gff = $blastjob->result_filestub . '.gff';
|
|
|
+
|
|
|
+ return $job;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Run BLAST (should be called from the command-line)
|
|
|
+ *
|
|
|
+ * @param $program
|
|
|
+ * Which BLAST program to run (ie: 'blastn', 'tblastn', tblastx', 'blastp','blastx')
|
|
|
+ * @param $query
|
|
|
+ * The full path and filename of the query FASTA file
|
|
|
+ * @param $database
|
|
|
+ * The full path and filename prefix (excluding .nhr, .nin, .nsq, etc.)
|
|
|
+ * @param $output_filestub
|
|
|
+ * The filename (not including path) to give the results. Should not include file type suffix
|
|
|
+ * @param $options
|
|
|
+ * An array of additional option where the key is the name of the option used by
|
|
|
+ * BLAST (ie: 'num_alignments') and the value is relates to this particular
|
|
|
+ * BLAST job (ie: 250)
|
|
|
+ */
|
|
|
+ public function runBLASTTripalJob($program, $query, $database, $output_filestub, $options, $job_id = NULL) {
|
|
|
+ $output_file = $output_filestub . '.asn';
|
|
|
+ $output_file_xml = $output_filestub . '.xml';
|
|
|
+ $output_file_tsv = $output_filestub . '.tsv';
|
|
|
+ $output_file_html = $output_filestub . '.html';
|
|
|
+ $output_file_gff = $output_filestub . '.gff';
|
|
|
+
|
|
|
+ print "\nExecuting $program\n\n";
|
|
|
+ print "Query: $query\n";
|
|
|
+ print "Database: $database\n";
|
|
|
+ print "Results File: $output_file\n";
|
|
|
+
|
|
|
+ print "Options:\n";
|
|
|
+
|
|
|
+ // Allow administrators to use an absolute path for these commands.
|
|
|
+ // Defaults to using $PATH.
|
|
|
+ $blast_path = variable_get('blast_path', '');
|
|
|
+ $blast_threads = variable_get('blast_threads', 1);
|
|
|
+
|
|
|
+ // Strip the extension off the BLAST target
|
|
|
+ $suffix = ['.ndb', '.nhr', '.nin', '.not', '.nsq', '.ntf', '.nto',
|
|
|
+ '.pdb', '.phr', '.pin', '.pot', '.psq', '.ptf', '.pto'];
|
|
|
+ $database = str_replace($suffix, '', $database);
|
|
|
+
|
|
|
+ // Check that the database exists before trying to execute the job.
|
|
|
+ if (!(file_exists($database . '.nsq') OR file_exists($database . '.psq'))) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_ERROR,
|
|
|
+ "Unable to find the BLAST database (ie: @db). Please ensure you have supplied the absolute path not including the file format endings.",
|
|
|
+ ['@db' => $database],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+
|
|
|
+ // The BLAST executeable.
|
|
|
+ $program = $blast_path . $program;
|
|
|
+ if (!file_exists($program)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_ERROR,
|
|
|
+ "Unable to find the BLAST executable (ie: /usr/bin/blastn). This can be changed in the admin settings; you supplied: @command",
|
|
|
+ ['@command' => $program],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+
|
|
|
+ // The blast db formatter executable.
|
|
|
+ $blast_formatter_command = $blast_path . 'blast_formatter';
|
|
|
+ if (!file_exists($blast_formatter_command)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_ERROR,
|
|
|
+ "Unable to find the BLAST Formatter executable (ie: /usr/bin/blast_formatter). This can be changed in the admin settings; you supplied: @command",
|
|
|
+ ['@command' => $blast_formatter_command],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+
|
|
|
+ // Note: all variables are escaped (adds single quotes around their values) for security reasons.
|
|
|
+ $blast_cmd = escapeshellarg($program) . ' -query ' . escapeshellarg($query) . ' -db ' . escapeshellarg($database) . ' -out ' . escapeshellarg($output_file) . ' -outfmt=11';
|
|
|
+ if (!empty($options)) {
|
|
|
+ foreach ($options as $opt => $val) {
|
|
|
+ $val = trim($val);
|
|
|
+ if (!empty($val)) {
|
|
|
+ print "\t$opt: $val\n";
|
|
|
+ // We want to escape all the option values since they were supplied via
|
|
|
+ // user input. These values should also have been checked in the
|
|
|
+ // advanced form _validate functions but this adds an extra layer of
|
|
|
+ // protection.
|
|
|
+ $blast_cmd .= ' -' . escapeshellarg($opt) . ' ' . escapeshellarg($val);
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // Setting the value of threads by admin page
|
|
|
+ $blast_cmd .= ' -num_threads ' . escapeshellarg($blast_threads);
|
|
|
+
|
|
|
+ print "\nExecuting the following BLAST command:\n" . $blast_cmd . "\n";
|
|
|
+
|
|
|
+ system($blast_cmd);
|
|
|
+
|
|
|
+ if (!file_exists($output_file)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_ERROR,
|
|
|
+ "BLAST did not complete successfully as is implied by the lack of output file (%file). The command run was @command",
|
|
|
+ ['%file' => $output_file, '@command' => $blast_cmd],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+
|
|
|
+ print "\nGenerating additional download formats...\n";
|
|
|
+
|
|
|
+ print "\tXML\n";
|
|
|
+ $format_cmd = escapeshellarg($blast_formatter_command) . ' -archive ' . escapeshellarg($output_file) . ' -outfmt 5 -out ' . escapeshellarg($output_file_xml);
|
|
|
+ print "\t\tExecuting $format_cmd\n\n";
|
|
|
+ system($format_cmd);
|
|
|
+ if (!file_exists($output_file_xml)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_ERROR,
|
|
|
+ "Unable to convert BLAST ASN.1 archive to XML (%archive => %file).",
|
|
|
+ ['%archive' => $output_file, '%file' => $output_file_xml],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+ }
|
|
|
+
|
|
|
+ print "\tTab-delimited\n";
|
|
|
+ $format_cmd = escapeshellarg($blast_formatter_command) . ' -archive ' . escapeshellarg($output_file) . ' -outfmt 7 -out ' . escapeshellarg($output_file_tsv);
|
|
|
+ print "\t\tExecuting $format_cmd\n\n";
|
|
|
+ system($format_cmd);
|
|
|
+ if (!file_exists($output_file_tsv)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_WARNING,
|
|
|
+ "Unable to convert BLAST ASN.1 archive to Tabular Output (%archive => %file).",
|
|
|
+ ['%archive' => $output_file, '%file' => $output_file_tsv],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+ }
|
|
|
+
|
|
|
+ print "\tGFF\n";
|
|
|
+ convert_tsv2gff3($output_file_tsv,$output_file_gff);
|
|
|
+
|
|
|
+ if (!file_exists($output_file_gff)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_WARNING,
|
|
|
+ "Unable to convert BLAST Tabular Output to GFF Output (%archive => %file).",
|
|
|
+ ['%archive' => $output_file, '%file' => $output_file_gff],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+ }
|
|
|
+
|
|
|
+ print "\tHTML (includes alignments)\n";
|
|
|
+ $format_cmd = escapeshellarg($blast_formatter_command) . ' -archive ' . escapeshellarg($output_file) . ' -outfmt 0 -out ' . escapeshellarg($output_file_html) . ' -html';
|
|
|
+ print "\t\tExecuting $format_cmd\n\n";
|
|
|
+ system($format_cmd);
|
|
|
+ if (!file_exists($output_file_tsv)) {
|
|
|
+ tripal_report_error(
|
|
|
+ 'blast_ui',
|
|
|
+ TRIPAL_WARNING,
|
|
|
+ "Unable to convert BLAST ASN.1 archive to HTML Output (%archive => %file).",
|
|
|
+ ['%archive' => $output_file, '%file' => $output_file_html],
|
|
|
+ ['print' => TRUE]
|
|
|
+ );
|
|
|
+ }
|
|
|
+
|
|
|
+ print "\nDone!\n";
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * FASTA validating parser
|
|
|
+ *
|
|
|
+ * A sequence in FASTA format begins with a single-line description, followed
|
|
|
+ * by lines of sequence data.The description line is distinguished from the
|
|
|
+ * sequence data by a greater-than (">") symbol in the first column. The word
|
|
|
+ * following the ">" symbol is the identifier of the sequence, and the rest of
|
|
|
+ * the line is the description (both are optional). There should be no space
|
|
|
+ * between the ">" and the first letter of the identifier. The sequence ends
|
|
|
+ * if another line starting with a ">" appears which indicates the start of
|
|
|
+ * another sequence.
|
|
|
+ *
|
|
|
+ * @param $type
|
|
|
+ * The type of sequence to be validated (ie: either nucleotide or protein).
|
|
|
+ * @param $sequence
|
|
|
+ * A string of characters to be validated.
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * Return a boolean. 1 if the sequence does not pass the format valifation stage and 0 otherwise.
|
|
|
+ *
|
|
|
+ */
|
|
|
+ function validateFASTASequence($type, $sequence) {
|
|
|
+ //Includes IUPAC codes.
|
|
|
+ $fastaSeqRegEx = ($type == 'nucleotide')
|
|
|
+ ? '/^[ATCGNUKMBVSWDYRHatcgnukmbvswdyrh\[\/\]\s\n\r]*$/'
|
|
|
+ : '/^[acdefghiklmnpqrstvwyACDEFGHIKLMNPQRSTVWY\*\-\s\n\r]*$/';
|
|
|
+ $defRegEx = '/^>\S.*/';
|
|
|
+
|
|
|
+ // For each line of the sequence.
|
|
|
+ foreach (explode("\n", $sequence) as $line) {
|
|
|
+
|
|
|
+ // Is this a definition line?
|
|
|
+ if ($line[0] == '>') {
|
|
|
+ if (!preg_match($defRegEx, $line)) {
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ // Otherwise it's a sequence line
|
|
|
+ else {
|
|
|
+ if (!preg_match($fastaSeqRegEx, $line)) {
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ return TRUE;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Retrieve the regex to capture the Link-out Accession from the Hit Def.
|
|
|
+ *
|
|
|
+ * @param $nid
|
|
|
+ * The node ID of the BLAST database the hit is from.
|
|
|
+ * @param $options
|
|
|
+ * An array of options that can be passed to this function. Supported
|
|
|
+ * options include:
|
|
|
+ * -
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * A PHP regex for use with preg_match to cature the Link-out Accession.
|
|
|
+ */
|
|
|
+ public function getBLASTDBLinkoutRegex($node, $options = array()) {
|
|
|
+
|
|
|
+ if (empty($node->linkout->regex)) {
|
|
|
+ switch ($node->linkout->regex_type) {
|
|
|
+ case 'default':
|
|
|
+ $regex = '/^(\S+).*/';
|
|
|
+ break;
|
|
|
+ case 'genbank':
|
|
|
+ $regex = '/^gb\|([^\|])*\|.*/';
|
|
|
+ break;
|
|
|
+ case 'embl':
|
|
|
+ $regex = '/^embl\|([^\|])*\|.*/';
|
|
|
+ break;
|
|
|
+ case 'swissprot':
|
|
|
+ $regex = '/^sp\|([^\|])*\|.*/';
|
|
|
+ break;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $regex = $node->linkout->regex;
|
|
|
+ }
|
|
|
+
|
|
|
+ return $regex;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Return a list of recent blast jobs to be displayed to the user.
|
|
|
+ *
|
|
|
+ * @param $programs
|
|
|
+ * An array of blast programs you want jobs to be displayed for (ie: blastn, blastx, tblastn, blastp)
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * An array of recent jobs.
|
|
|
+ */
|
|
|
+ public function get_recent_blast_jobs($programs = array()) {
|
|
|
+
|
|
|
+ $filter_jobs = !empty($programs);
|
|
|
+
|
|
|
+ // Retrieve any recent jobs from the session variable.
|
|
|
+ if (isset($_SESSION['blast_jobs'])) {
|
|
|
+
|
|
|
+ $jobs = array();
|
|
|
+ foreach ($_SESSION['blast_jobs'] as $job_secret) {
|
|
|
+ $add = TRUE;
|
|
|
+
|
|
|
+ $job_id = blast_ui_reveal_secret($job_secret);
|
|
|
+ if ($job = get_BLAST_job($job_id)) {
|
|
|
+
|
|
|
+ // @TODO: Check that the results are still available.
|
|
|
+ // This is meant to replace the arbitrary only show jobs executed less than 48 hrs ago.
|
|
|
+
|
|
|
+ // Remove jobs from the list that are not of the correct program.
|
|
|
+ if ($filter_jobs AND !in_array($job->program, $programs)) {
|
|
|
+ $add = FALSE;
|
|
|
+ }
|
|
|
+
|
|
|
+ if ($add) {
|
|
|
+ $job->query_summary = format_query_headers($job->files->query);
|
|
|
+ $jobs[] = $job;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ return $jobs;
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ return array();
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Retrieve the number of recent jobs.
|
|
|
+ */
|
|
|
+ public function getNumberOfRecentJobs() {
|
|
|
+ if (isset($_SESSION['blast_jobs'])) {
|
|
|
+ return sizeof($_SESSION['blast_jobs']);
|
|
|
+ }
|
|
|
+
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Summarize a fasta file based on it's headers.
|
|
|
+ *
|
|
|
+ * @param $file
|
|
|
+ * The full path to the FASTA file.
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * A string describing the number of sequences and often including the first query header.
|
|
|
+ */
|
|
|
+ public function format_query_headers($file) {
|
|
|
+
|
|
|
+ $headers = [];
|
|
|
+ exec('grep ">" ' . escapeshellarg($file), $headers);
|
|
|
+
|
|
|
+ // Easiest case: if there is only one query header then show it.
|
|
|
+ if (sizeof($headers) == 1 AND isset($headers[0])) {
|
|
|
+ return ltrim($headers[0], '>');
|
|
|
+ }
|
|
|
+ // If we have at least one header then show that along with the count of queries.
|
|
|
+ elseif (isset($headers[0])) {
|
|
|
+ return sizeof($headers) . ' queries including "' . ltrim($headers[0], '>') . '"';
|
|
|
+ }
|
|
|
+ // If they provided a sequence with no header.
|
|
|
+ elseif (empty($headers)) {
|
|
|
+ return 'Unnamed Query';
|
|
|
+ }
|
|
|
+ // At the very least show the count of queries.
|
|
|
+ else {
|
|
|
+ return sizeof($headers) . ' queries';
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Sort recent blast jobs by the date they were submitted.
|
|
|
+ * Ascending is in order of submission.
|
|
|
+ *
|
|
|
+ * THIS FUNCTION SHOULD BY USED BY USORT().
|
|
|
+ */
|
|
|
+ public function sortBLASTJobsByDateSubmittedAsc($a, $b) {
|
|
|
+ return ($a->date_submitted - $b->date_submitted);
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Sort recent blast jobs by the date they were submitted.
|
|
|
+ * Descending is most recent first.
|
|
|
+ *
|
|
|
+ * THIS FUNCTION SHOULD BY USED BY USORT().
|
|
|
+ */
|
|
|
+ public function sortBLASTJobsByDateSubmittedDesc($a, $b) {
|
|
|
+ return ($b->date_submitted - $a->date_submitted);
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Generate an image of HSPs for a given hit.
|
|
|
+ *
|
|
|
+ * history:
|
|
|
+ * 09/23/10 Carson created
|
|
|
+ * 04/16/12 eksc adapted into POPcorn code
|
|
|
+ * 03/12/15 deepak Adapted code into Tripal BLAST
|
|
|
+ * 10/23/15 lacey Fixed deepak's code to be suitable for Drupal.
|
|
|
+ *
|
|
|
+ * @param $acc
|
|
|
+ * target name
|
|
|
+ * @param $name
|
|
|
+ * query name, false if none
|
|
|
+ * @param $tsize
|
|
|
+ * target size
|
|
|
+ * @param $qsize
|
|
|
+ * query size
|
|
|
+ * @param $hits
|
|
|
+ * each hit represented in URL as: targetstart_targetend_hspstart_hspend;
|
|
|
+ * @param $score
|
|
|
+ * score for each hit
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * A base64 encoded image representing the hit information.
|
|
|
+ */
|
|
|
+ public function generate_blast_hit_image($acc = '', $scores, $hits, $tsize, $qsize, $name, $hit_name) {
|
|
|
+ $tok = strtok($hits, ";");
|
|
|
+ $b_hits = [];
|
|
|
+
|
|
|
+ while ($tok !== false) {
|
|
|
+ $b_hits[] = $tok;
|
|
|
+ $tok = strtok(";");
|
|
|
+ }
|
|
|
+
|
|
|
+ // extract score information from score param
|
|
|
+ $tokscr = strtok($scores, ";");
|
|
|
+ $b_scores = [];
|
|
|
+
|
|
|
+ while ($tokscr !== false) {
|
|
|
+ $b_scores[] = $tokscr;
|
|
|
+ $tokscr = strtok(";");
|
|
|
+ }
|
|
|
+
|
|
|
+ // image measurements
|
|
|
+ $height = 200 + (count($b_hits) * 16);
|
|
|
+ $width = 520;
|
|
|
+
|
|
|
+ $img = imagecreatetruecolor($width, $height);
|
|
|
+
|
|
|
+ $white = imagecolorallocate($img, 255, 255, 255);
|
|
|
+ $black = imagecolorallocate($img, 0, 0, 0);
|
|
|
+ $darkgray = imagecolorallocate($img, 100, 100, 100);
|
|
|
+ $strong = imagecolorallocatealpha($img, 202, 0, 0, 15);
|
|
|
+ $moderate = imagecolorallocatealpha($img, 204, 102, 0, 20);
|
|
|
+ $present = imagecolorallocatealpha($img, 204, 204, 0, 35);
|
|
|
+ $weak = imagecolorallocatealpha($img, 102, 204, 0, 50);
|
|
|
+ $gray = imagecolorallocate($img, 190, 190, 190);
|
|
|
+ $lightgray = $white; //imagecolorallocate($img, 230, 230, 230);
|
|
|
+
|
|
|
+ imagefill($img, 0, 0, $lightgray);
|
|
|
+
|
|
|
+ // Target coordinates
|
|
|
+ $maxlength = 300;
|
|
|
+ $t_length = ($tsize > $qsize)
|
|
|
+ ? $maxlength : $maxlength - 50;
|
|
|
+ $q_length = ($qsize > $tsize)
|
|
|
+ ? $maxlength : $maxlength - 50;
|
|
|
+
|
|
|
+ $tnormal = $t_length / $tsize;
|
|
|
+ $qnormal = $q_length / $qsize;
|
|
|
+
|
|
|
+ $t_ystart = 30;
|
|
|
+ $t_yend = $t_ystart + 20;
|
|
|
+
|
|
|
+ $t_xstart = 50;
|
|
|
+ $t_xend = $t_xstart + $t_length;
|
|
|
+ $t_center = $t_xstart + ($t_length / 2);
|
|
|
+
|
|
|
+ // Target labels
|
|
|
+ $warn = '"'. $hit_name . '"';
|
|
|
+ imagestring($img, 5, $t_xstart, $t_ystart-20, $acc.$warn, $black);
|
|
|
+ imagestring($img, 3, 5, $t_ystart+2, "Target", $black);
|
|
|
+
|
|
|
+ // Draw bar representing target
|
|
|
+ imagefilledrectangle($img, $t_xstart, $t_ystart, $t_xend, $t_yend, $gray);
|
|
|
+ imagerectangle($img, $t_xstart, $t_ystart, $t_xend, $t_yend, $darkgray);
|
|
|
+
|
|
|
+ // query coordinates
|
|
|
+ $q_maxheight = 250;
|
|
|
+ $q_ystart = $t_yend + 100;
|
|
|
+ $q_yend = $q_ystart + 20;
|
|
|
+
|
|
|
+ $q_xstart = $t_center - $q_length / 2;
|
|
|
+ $q_xend = $q_xstart + $q_length;
|
|
|
+
|
|
|
+ $q_center = ($q_xend + $q_xstart) / 2;
|
|
|
+ $q_xwidth = $q_xend - $q_xstart;
|
|
|
+
|
|
|
+ // Query labels
|
|
|
+ imagestring($img, 5, $q_xstart, $q_yend+2, $name, $black);
|
|
|
+ imagestring($img, 3, $q_xstart, $q_ystart+2, 'Query', $black);
|
|
|
+
|
|
|
+ // Draw bar representing query
|
|
|
+ imagefilledrectangle($img, $q_xstart, $q_ystart, $q_xend, $q_yend, $gray);
|
|
|
+ imagerectangle($img ,$q_xstart, $q_ystart, $q_xend, $q_yend, $darkgray);
|
|
|
+
|
|
|
+ // HSP bars will start here
|
|
|
+ $hsp_bary = $q_yend + 20;
|
|
|
+
|
|
|
+ // Draw solids for HSP alignments
|
|
|
+ for ($ii=count($b_hits)-1; $ii>=0; $ii--) {
|
|
|
+ // alignment
|
|
|
+
|
|
|
+ $cur_hit = $b_hits[$ii];
|
|
|
+ $cur_score = intval($b_scores[$ii]);
|
|
|
+
|
|
|
+ // set color according to score
|
|
|
+ $cur_color = $darkgray;
|
|
|
+ if ($cur_score > 200) {
|
|
|
+ $cur_color = $strong;
|
|
|
+ }
|
|
|
+ else if ($cur_score > 80 && $cur_score <= 200) {
|
|
|
+ $cur_color = $moderate;
|
|
|
+ }
|
|
|
+ else if ($cur_score > 50 && $cur_score <= 80) {
|
|
|
+ $cur_color = $present;
|
|
|
+ }
|
|
|
+ else if ($cur_score > 40 && $cur_score <= 50) {
|
|
|
+ $cur_color = $weak;
|
|
|
+ }
|
|
|
+
|
|
|
+ $t_start = $tnormal * intval(strtok($cur_hit, "_")) + $t_xstart;
|
|
|
+ $t_end = $tnormal * intval(strtok("_")) + $t_xstart;
|
|
|
+ $q_start = $qnormal * intval(strtok("_")) + $q_xstart;
|
|
|
+ $q_end = $qnormal * intval(strtok("_")) + $q_xstart;
|
|
|
+
|
|
|
+ $hit1_array = [$t_start, $t_yend, $t_end, $t_yend, $q_end,
|
|
|
+ $q_ystart, $q_start, $q_ystart];
|
|
|
+
|
|
|
+ // HSP coords
|
|
|
+ imagefilledpolygon($img, $hit1_array, 4, $cur_color);
|
|
|
+
|
|
|
+ }//each hit
|
|
|
+
|
|
|
+ // Draw lines over fills for HSP alignments
|
|
|
+ for ($ii=0; $ii<count($b_hits); $ii++) {
|
|
|
+ // alignment
|
|
|
+
|
|
|
+ $cur_hit = $b_hits[$ii];
|
|
|
+ $t_start = $tnormal * intval(strtok($cur_hit, "_")) + $t_xstart;
|
|
|
+ $t_end = $tnormal * intval(strtok("_")) + $t_xstart;
|
|
|
+ $q_start = $qnormal * intval(strtok("_")) + $q_xstart;
|
|
|
+ $q_end = $qnormal * intval(strtok("_")) + $q_xstart;
|
|
|
+
|
|
|
+ $hit1_array = array($t_start, $t_yend, $t_end, $t_yend, $q_end, $q_ystart,
|
|
|
+ $q_start, $q_ystart,);
|
|
|
+
|
|
|
+ imagerectangle($img, $t_start, $t_ystart, $t_end, $t_yend, $black);
|
|
|
+ imagerectangle($img, $q_start, $q_ystart, $q_end, $q_yend, $black);
|
|
|
+ imagepolygon ($img, $hit1_array, 4, $black);
|
|
|
+
|
|
|
+ // show HSP
|
|
|
+
|
|
|
+ imagestring($img, 3, 2, $hsp_bary, ($acc ."HSP" . ($ii + 1)), $black);
|
|
|
+
|
|
|
+ $cur_score = intval($b_scores[$ii]);
|
|
|
+
|
|
|
+ // set color according to score
|
|
|
+ $cur_color = $darkgray;
|
|
|
+ if ($cur_score > 200) {
|
|
|
+ $cur_color = $strong;
|
|
|
+ }
|
|
|
+ else if ($cur_score > 80 && $cur_score <= 200) {
|
|
|
+ $cur_color = $moderate;
|
|
|
+ }
|
|
|
+ else if ($cur_score > 50 && $cur_score <= 80) {
|
|
|
+ $cur_color = $present;
|
|
|
+ }
|
|
|
+ else if ($cur_score > 40 && $cur_score <= 50) {
|
|
|
+ $cur_color = $weak;
|
|
|
+ }
|
|
|
+
|
|
|
+ imagefilledrectangle($img, $q_start, $hsp_bary, $q_end, $hsp_bary+10, $cur_color);
|
|
|
+ $hsp_bary += 15;
|
|
|
+ }//each hit
|
|
|
+
|
|
|
+ // Draw the key
|
|
|
+
|
|
|
+ $xchart = 390;
|
|
|
+ $ychart = 10;
|
|
|
+ $fontsize = 4;
|
|
|
+ $yinc = 20;
|
|
|
+ $ywidth = 7;
|
|
|
+ $xinc = 10;
|
|
|
+
|
|
|
+ imagestring($img, 5, $xchart, $ychart - 5, "Bit Scores", $black);
|
|
|
+
|
|
|
+ imagestring($img, $fontsize, $xchart + $yinc + $xinc,$ychart + ($yinc * 1) + $ywidth, ">= 200" , $black);
|
|
|
+ imagestring($img, $fontsize, $xchart + $yinc + $xinc,$ychart + ($yinc * 2) + $ywidth, "80 - 200" , $black);
|
|
|
+ imagestring($img, $fontsize, $xchart + $yinc + $xinc,$ychart + ($yinc * 3) + $ywidth, "50 - 80" , $black);
|
|
|
+ imagestring($img, $fontsize, $xchart + $yinc + $xinc,$ychart + ($yinc * 4) + $ywidth, "40 - 50" , $black);
|
|
|
+ imagestring($img, $fontsize, $xchart + $yinc + $xinc,$ychart + ($yinc * 5) + $ywidth, "< 40" , $black);
|
|
|
+
|
|
|
+ imagefilledRectangle($img, $xchart, $ychart + ($yinc * 1) + $xinc, $xchart + $yinc, $ychart + ($yinc * 2), $strong);
|
|
|
+ imagefilledRectangle($img, $xchart, $ychart + ($yinc * 2) + $xinc, $xchart + $yinc, $ychart + ($yinc * 3), $moderate);
|
|
|
+ imagefilledRectangle($img, $xchart, $ychart + ($yinc * 3) + $xinc, $xchart + $yinc, $ychart + ($yinc * 4), $present);
|
|
|
+ imagefilledRectangle($img, $xchart, $ychart + ($yinc * 4) + $xinc, $xchart + $yinc, $ychart + ($yinc * 5), $weak);
|
|
|
+ imagefilledRectangle($img, $xchart, $ychart + ($yinc * 5) + $xinc, $xchart + $yinc, $ychart + ($yinc * 6), $darkgray);
|
|
|
+
|
|
|
+ // Now, we have a completed image resource and need to change it to an actual image
|
|
|
+ // that can be displayed. This is done using imagepng() but unfortuatly that function
|
|
|
+ // either saves the image to a file or outputs it directly to the screen. Thus, we use
|
|
|
+ // the following code to capture it and base64 encode it.
|
|
|
+ ob_start(); // Start buffering the output
|
|
|
+ imagepng($img, null, 0, PNG_NO_FILTER);
|
|
|
+ $b64_img = base64_encode(ob_get_contents()); // Get what we've just outputted and base64 it
|
|
|
+ imagedestroy($img);
|
|
|
+ ob_end_clean();
|
|
|
+
|
|
|
+ return $b64_img;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Convert tsv blast output to gff output file.
|
|
|
+ *
|
|
|
+ * Created by Sofia Robb
|
|
|
+ * 09/15/2016
|
|
|
+ * counter bugfix 10/27/2016
|
|
|
+ *
|
|
|
+ * The subject (hit) will be the source feature.
|
|
|
+ * The query will be the target.
|
|
|
+ *
|
|
|
+ * @todo: find a more efficient way since currently the first loop stores all the blast
|
|
|
+ * results into an array and then the second loop prints them.
|
|
|
+ *
|
|
|
+ * @param $blast_tsv
|
|
|
+ * The name of the blast tsv output file.
|
|
|
+ * @param $blast_gff
|
|
|
+ * The name of the blast gff output file.
|
|
|
+ */
|
|
|
+ public function convertTsvToGff3($blast_tsv, $blast_gff){
|
|
|
+
|
|
|
+ // Open a new file for writting the gff.
|
|
|
+ $gff = fopen($blast_gff,"w");
|
|
|
+ fwrite($gff,"##gff-version 3\n");
|
|
|
+
|
|
|
+ // Open the TSV file to read from.
|
|
|
+ $tsv = fopen($blast_tsv, "r") or die("Unable to open tsv file!");
|
|
|
+
|
|
|
+ // For each line in the TSV file...
|
|
|
+ // Need to go thru each line of tsv to find the first and last hsp of a hit.
|
|
|
+ $last_s = NULL;
|
|
|
+ $hsp = NULL;
|
|
|
+ $HitResult = [];
|
|
|
+
|
|
|
+ while(!feof($tsv)) {
|
|
|
+ $line = fgets($tsv);
|
|
|
+ $line = rtrim($line);
|
|
|
+
|
|
|
+ // Skip the line if it's empty.
|
|
|
+ if (preg_match('/^#/',$line) or preg_match('/^\s*$/',$line)){
|
|
|
+ continue;
|
|
|
+ }
|
|
|
+
|
|
|
+ ## for keeping track of new queries and hits
|
|
|
+
|
|
|
+ // Each line has the following parts:
|
|
|
+ // 0: query id,
|
|
|
+ // 1: subject id,
|
|
|
+ // 2: % identity,
|
|
|
+ // 3: alignment length,
|
|
|
+ // 4: mismatches,
|
|
|
+ // 5: gap opens,
|
|
|
+ // 6: q. start,
|
|
|
+ // 7: q. end,
|
|
|
+ // 8: s. start,
|
|
|
+ // 9: s. end,
|
|
|
+ // 10: evalue,
|
|
|
+ // 11: bit score
|
|
|
+ $parts = preg_split('/\t/', $line);
|
|
|
+
|
|
|
+ // Assign the important parts of the line to readable variables.
|
|
|
+ $s = $parts[1];
|
|
|
+ $q = $parts[0];
|
|
|
+ $ss = $parts[8];
|
|
|
+ $se = $parts[9];
|
|
|
+ $qs = $parts[6];
|
|
|
+ $qe = $parts[7];
|
|
|
+ $e = $parts[10];
|
|
|
+
|
|
|
+ // if this is a new hit print the last and
|
|
|
+ // empty the $HitResult array and
|
|
|
+ // reset hsp counter
|
|
|
+ if ($last_s != NULL and $s != $last_s ) {
|
|
|
+ $this->printGFFParentChildren($gff,$HitResult);
|
|
|
+ $HitResult = array();
|
|
|
+ $hsp=0;
|
|
|
+ }
|
|
|
+
|
|
|
+ // every line is a new hsp
|
|
|
+ $hsp++;
|
|
|
+
|
|
|
+ // determine query strand to use in match_part line, no need to store, just print
|
|
|
+ $q_strand = '+';
|
|
|
+ if ($qs > $qe) {
|
|
|
+ list($qs,$qe) = array($qe,$qs);
|
|
|
+ $q_strand = '-';
|
|
|
+ }
|
|
|
+
|
|
|
+ // determine subject (hit) strand to use in match line, needs to be stored
|
|
|
+ $HitResult["$s,$q"]['strand']='+';
|
|
|
+ list($start,$end) = array($ss,$se);
|
|
|
+ if($ss > $se) {
|
|
|
+ list($start,$end) = array($se,$ss);
|
|
|
+ $HitResult["$s,$q"]['strand']='-';
|
|
|
+ }
|
|
|
+
|
|
|
+ // store smallest start
|
|
|
+ if (!array_key_exists('SS',$HitResult["$s,$q"]) or $ss < $HitResult["$s,$q"]['SS']) {
|
|
|
+ $HitResult["$s,$q"]['SS'] = $ss;
|
|
|
+ }
|
|
|
+
|
|
|
+ // store largest end
|
|
|
+ if (!array_key_exists('SE',$HitResult["$s,$q"]) or $se > $HitResult["$s,$q"]['SE']) {
|
|
|
+ $HitResult["$s,$q"]['SE'] = $se;
|
|
|
+ }
|
|
|
+
|
|
|
+ // store best evalue
|
|
|
+ if (!array_key_exists('E',$HitResult["$s,$q"]) or $e < $HitResult["$s,$q"]['E']) {
|
|
|
+ $HitResult["$s,$q"]['E'] = $e;
|
|
|
+ }
|
|
|
+
|
|
|
+ // generate the match_part line for each hsp
|
|
|
+ $HitResult["$s,$q"]['HSPs'][] = join("\t", [$s, "BLASTRESULT" , "match_part" , $start , $end , $e , $HitResult["$s,$q"]['strand'] , '.' , "ID=$s.$q.$hsp;Parent=$s.$q;Target=$q $qs $qe $q_strand"]);
|
|
|
+ $last_s = $s;
|
|
|
+ } // end tsv file while
|
|
|
+
|
|
|
+ // print hit and hsp for the last hit
|
|
|
+ $this->printGFFParentChildren($gff, $HitResult);
|
|
|
+
|
|
|
+ // Close the files.
|
|
|
+ fclose($tsv);
|
|
|
+ fclose($gff);
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Prints the GFF parent feature and all of its children features
|
|
|
+ *
|
|
|
+ * @param $blast_feature_array
|
|
|
+ * an array of the all the child features which is used to generate the smallest and largest coordinates for the parent
|
|
|
+ */
|
|
|
+ public function printGFFParentChildren($gff, $blast_feature_array){
|
|
|
+ foreach ($blast_feature_array as $sq => $value ) {
|
|
|
+ list ($s,$q) = preg_split('/,/', $sq);
|
|
|
+ $evalue = $blast_feature_array["$s,$q"]['E'];
|
|
|
+ $parent = join("\t", array($s, "BLASTRESULT", "match", $blast_feature_array["$s,$q"]['SS'], $blast_feature_array["$s,$q"]['SE'], $blast_feature_array["$s,$q"]['E'], $blast_feature_array["$s,$q"]['strand'], '.' , "ID=$s.$q;Name=$q($evalue)")) . "\n";
|
|
|
+ $child = join("\n", $blast_feature_array["$s,$q"]['HSPs']) . "\n";
|
|
|
+
|
|
|
+ fwrite($gff,$parent);
|
|
|
+ fwrite($gff,$child);
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Get text from cvitjs conf file, if possible.
|
|
|
+ *
|
|
|
+ * @param $genome_target
|
|
|
+ * The section of the config to return. Should consist of "data."+[blastdb name].
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * A string containing the entire contents of the cvitjs configuration file. FALSE otherwise.
|
|
|
+ */
|
|
|
+ public function BLASTUIGetCivitConfText($genome_target = FALSE) {
|
|
|
+
|
|
|
+ // Retrieve the full path and filename of the conf.
|
|
|
+ $cvit_conf = $this->BLASTUIGetCivitConf();
|
|
|
+ if ($cvit_conf) {
|
|
|
+
|
|
|
+ // Retrieve the contents of the file.
|
|
|
+ $contents = '';
|
|
|
+ if (file_exists($cvit_conf)) {
|
|
|
+ $contents = file_get_contents($cvit_conf);
|
|
|
+ }
|
|
|
+
|
|
|
+ // If no genome target was provided then return the full file.
|
|
|
+ if ($contents && $genome_target == FALSE) {
|
|
|
+ return $contents;
|
|
|
+ }
|
|
|
+
|
|
|
+ // If a genome target was provided, then only return that section.
|
|
|
+ if ($genome_target) {
|
|
|
+ $section = array();
|
|
|
+ $in_section = FALSE;
|
|
|
+
|
|
|
+ // For each line of the configuration file...
|
|
|
+ $section_header = '['.$genome_target.']';
|
|
|
+ $lines = preg_split('/\r\n|\n|\r/', trim($contents));
|
|
|
+ foreach($lines as $l) {
|
|
|
+
|
|
|
+ // Are we in the section for this genome target?
|
|
|
+ if (trim($l) == $section_header) {
|
|
|
+ $in_section = TRUE; }
|
|
|
+
|
|
|
+ // Id so and we haven't fallen out of it through an empty line,
|
|
|
+ // then add it to saved section for returning.
|
|
|
+ if ($in_section) {
|
|
|
+ if (trim($l) == '') { break; }
|
|
|
+ $section[] = trim($l);
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // If we found the section, then return it ;-).
|
|
|
+ if (!empty($section)) {
|
|
|
+ return implode("\n", $section);
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ return false;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * Get path to cvitjs conf file.
|
|
|
+ *
|
|
|
+ * @return
|
|
|
+ * The path to the CViTjs codebase.
|
|
|
+ */
|
|
|
+ public function BLASTUIGetCivitConf($cvitjs_location = NULL) {
|
|
|
+ if (!$cvitjs_location) {
|
|
|
+ $cvitjs_location = libraries_get_path('cvitjs') . DIRECTORY_SEPARATOR;
|
|
|
+ }
|
|
|
+
|
|
|
+ $cvit_conf_path = $cvitjs_location . 'cvit.conf';
|
|
|
+
|
|
|
+ return $cvit_conf_path;
|
|
|
+ }
|
|
|
+}
|
Reynold Tan