function fetch_sequence($id)
{
$genbank_sequence = null;
// Query URL
$url = 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nucleotide&id=' . $id . '&rettype=gb&retmode=xml';
$xml = get($url);
//echo $xml;
if ($xml != '') {
$xp = new XsltProcessor();
$xsl = new DomDocument();
$xsl->load('xml2json.xslt');
$xp->importStylesheet($xsl);
$dom = new DOMDocument();
$dom->loadXML($xml);
$xpath = new DOMXPath($dom);
$json = $xp->transformToXML($dom);
//echo $json;
// fix "-" in variable names
$json = str_replace('"GBSeq_feature-table"', '"GBSeq_feature_table"', $json);
$json = str_replace('"GBSeq_primary-accession"', '"GBSeq_primary_accession"', $json);
$json = str_replace('"GBSeq_other-seqids"', '"GBSeq_other_seqids"', $json);
$json = str_replace('"GBSeq_update-date"', '"GBSeq_update_date"', $json);
$json = str_replace('"GBSeq_create-date"', '"GBSeq_create_date"', $json);
$json = str_replace('"GBSeq_accession-version"', '"GBSeq_accession_version"', $json);
$sequences = json_decode($json);
//print_r($sequences);
foreach ($sequences->GBSet as $GBSet) {
$genbank_sequence = new stdclass();
$genbank_sequence->accession = $GBSet->GBSeq_primary_accession;
$genbank_sequence->accession_version = $GBSet->GBSeq_accession_version;
$genbank_sequence->organism = $GBSet->GBSeq_organism;
$genbank_sequence->definition = $GBSet->GBSeq_definition;
$genbank_sequence->moltype = $GBSet->GBSeq_moltype;
// dates
if (false != strtotime($GBSet->GBSeq_update_date)) {
$genbank_sequence->updated = date("Y-m-d", strtotime($GBSet->GBSeq_update_date));
}
if (false != strtotime($GBSet->GBSeq_create_date)) {
$genbank_sequence->created = date("Y-m-d", strtotime($GBSet->GBSeq_create_date));
}
// keywords
$genbank_sequence->barcode = false;
if ($GBSet->GBSeq_keywords) {
foreach ($GBSet->GBSeq_keywords as $keyword) {
if ($keyword = 'BARCODE') {
$genbank_sequence->barcode = true;
}
}
}
foreach ($GBSet->GBSeq_other_seqids as $seqids) {
if (preg_match('/gi\\|(?<gi>\\d+)$/', $seqids, $m)) {
$genbank_sequence->gi = (int) $m['gi'];
}
}
$genbank_sequence->references = array();
foreach ($GBSet->GBSeq_references as $GBReference) {
$reference = new stdclass();
$reference->title = $GBReference->GBReference_title;
$reference->citation = $GBReference->GBReference_journal;
if (isset($GBReference->GBReference_authors)) {
foreach ($GBReference->GBReference_authors as $a) {
$parts = parse_name($a);
$author = new stdClass();
$author->name = $a;
if (isset($parts['last'])) {
$author->lastname = $parts['last'];
}
if (isset($parts['first'])) {
$author->forename = $parts['first'];
if (array_key_exists('middle', $parts)) {
$author->forename .= ' ' . $parts['middle'];
}
}
$reference->author[] = $author;
}
}
if (preg_match('/(?<journal>.*)\\s+(?<volume>\\d+)(\\s+\\((?<issue>.*)\\))?,\\s+(?<spage>\\d+)-(?<epage>\\d+)\\s+\\((?<year>[0-9]{4})\\)/', $reference->citation, $m)) {
$reference->journal = new stdclass();
$reference->journal->name = $m['journal'];
$reference->journal->volume = $m['volume'];
if ($m['issue'] != '') {
$reference->journal->issue = $m['issue'];
}
$reference->journal->pages = $m['spage'];
if ($m['epage'] != '') {
$reference->journal->pages .= '--' . $m['epage'];
}
}
if (isset($GBReference->GBReference_pubmed)) {
$identifier = new stdclass();
$identifier->type = 'pmid';
$identifier->id = (int) $GBReference->GBReference_pubmed;
$reference->identifier[] = $identifier;
}
if (isset($GBReference->GBReference_xref)) {
if ($GBReference->GBReference_xref->GBXref->GBXref_dbname == 'doi') {
$identifier = new stdclass();
$identifier->type = 'doi';
$identifier->id = $GBReference->GBReference_xref->GBXref->GBXref_id;
$reference->identifier[] = $identifier;
}
}
$genbank_sequence->references[] = $reference;
}
foreach ($GBSet->GBSeq_feature_table as $feature_table) {
switch ($feature_table->GBFeature_key) {
case 'source':
if (!isset($genbank_sequence->source)) {
$genbank_sequence->source = new stdclass();
}
foreach ($feature_table->GBFeature_quals as $feature_quals) {
switch ($feature_quals->GBQualifier_name) {
case 'db_xref':
$genbank_sequence->source->tax_id = (int) str_replace("taxon:", '', $feature_quals->GBQualifier_value);
break;
case 'collection_date':
$genbank_sequence->source->collection_date = $feature_quals->GBQualifier_value;
break;
case 'collected_by':
$genbank_sequence->source->collected_by = $feature_quals->GBQualifier_value;
break;
case 'country':
$genbank_sequence->source->country = $feature_quals->GBQualifier_value;
break;
case 'host':
$genbank_sequence->source->host = $feature_quals->GBQualifier_value;
break;
case 'locality':
$genbank_sequence->source->locality = $feature_quals->GBQualifier_value;
break;
case 'isolation_source':
$genbank_sequence->source->isolation_source = $feature_quals->GBQualifier_value;
break;
case 'isolate':
$genbank_sequence->source->isolate = $feature_quals->GBQualifier_value;
break;
case 'lat_lon':
$genbank_sequence->source->lat_lon = $feature_quals->GBQualifier_value;
break;
case 'mol_type':
$genbank_sequence->source->mol_type = $feature_quals->GBQualifier_value;
break;
case 'organelle':
$genbank_sequence->source->organelle = $feature_quals->GBQualifier_value;
break;
case 'specimen_voucher':
$genbank_sequence->source->specimen_voucher = $feature_quals->GBQualifier_value;
break;
default:
break;
}
}
if (isset($genbank_sequence->source->lat_lon)) {
process_lat_lon($genbank_sequence);
}
process_locality($genbank_sequence);
break;
default:
break;
}
}
if (isset($genbank_sequence->source->latitude)) {
$genbank_sequence->source->geometry = new stdclass();
$genbank_sequence->source->geometry->type = "Point";
$genbank_sequence->source->geometry->coordinates = array((double) $genbank_sequence->source->longitude, (double) $genbank_sequence->source->latitude);
}
}
}
return $genbank_sequence;
}
function fetch_sequences($ids)
{
$hits = new stdclass();
$hits->ids = $ids;
$hits->sequences = array();
$hits->geometry = new stdclass();
$hits->geometry->type = "MultiPoint";
$hits->geometry->coordinates = array();
// Query URL
$url = 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nucleotide&id=' . join(",", $ids) . '&rettype=gb&retmode=xml';
$xml = get($url);
if ($xml != '') {
$xp = new XsltProcessor();
$xsl = new DomDocument();
$xsl->load('xml2json.xslt');
$xp->importStylesheet($xsl);
$dom = new DOMDocument();
$dom->loadXML($xml);
$xpath = new DOMXPath($dom);
$json = $xp->transformToXML($dom);
// fix "-" in variable names
$json = str_replace('"GBSeq_feature-table"', '"GBSeq_feature_table"', $json);
$json = str_replace('"GBSeq_primary-accession"', '"GBSeq_primary_accession"', $json);
$json = str_replace('"GBSeq_other-seqids"', '"GBSeq_other_seqids"', $json);
$json = str_replace('"GBSeq_create-date"', '"GBSeq_create_date"', $json);
$json = str_replace('"GBSeq_update-date"', '"GBSeq_update_date"', $json);
$sequences = json_decode($json);
foreach ($sequences->GBSet as $GBSet) {
$genbank_sequence = new stdclass();
$genbank_sequence->accession = $GBSet->GBSeq_primary_accession;
$genbank_sequence->organism = $GBSet->GBSeq_organism;
$genbank_sequence->created = format_date($GBSet->GBSeq_create_date);
$genbank_sequence->updated = format_date($GBSet->GBSeq_update_date);
foreach ($GBSet->GBSeq_other_seqids as $seqids) {
if (preg_match('/gi\\|(?<gi>\\d+)$/', $seqids, $m)) {
$genbank_sequence->gi = $m['gi'];
}
}
$genbank_sequence->references = array();
foreach ($GBSet->GBSeq_references as $GBReference) {
$reference = new stdclass();
$reference->title = $GBReference->GBReference_title;
$reference->citation = $GBReference->GBReference_journal;
$reference->authors = array();
if (isset($GBReference->GBReference_authors)) {
foreach ($GBReference->GBReference_authors as $author) {
$reference->authors[] = $author;
}
}
if (isset($GBReference->GBReference_pubmed)) {
if (!isset($reference->identifiers)) {
$reference->identifiers = new stdclass();
}
$reference->identifiers->pmid = $GBReference->GBReference_pubmed;
}
if (isset($GBReference->GBReference_xref)) {
if ($GBReference->GBReference_xref->GBXref->GBXref_dbname == 'doi') {
if (!isset($reference->identifiers)) {
$reference->identifiers = new stdclass();
}
$reference->identifiers->doi = $GBReference->GBReference_xref->GBXref->GBXref_id;
}
}
$genbank_sequence->references[] = $reference;
}
foreach ($GBSet->GBSeq_feature_table as $feature_table) {
switch ($feature_table->GBFeature_key) {
case 'source':
foreach ($feature_table->GBFeature_quals as $feature_quals) {
switch ($feature_quals->GBQualifier_name) {
case 'db_xref':
$genbank_sequence->source->tax_id = str_replace("taxon:", '', $feature_quals->GBQualifier_value);
break;
case 'country':
$genbank_sequence->source->country = $feature_quals->GBQualifier_value;
break;
case 'host':
$genbank_sequence->source->host = $feature_quals->GBQualifier_value;
break;
case 'locality':
$genbank_sequence->source->locality = $feature_quals->GBQualifier_value;
break;
case 'isolation_source':
$genbank_sequence->source->isolation_source = $feature_quals->GBQualifier_value;
break;
case 'isolate':
$genbank_sequence->source->isolate = $feature_quals->GBQualifier_value;
break;
case 'lat_lon':
$genbank_sequence->source->lat_lon = $feature_quals->GBQualifier_value;
break;
case 'specimen_voucher':
$genbank_sequence->source->specimen_voucher = $feature_quals->GBQualifier_value;
break;
default:
break;
}
}
if (isset($genbank_sequence->source->lat_lon)) {
process_lat_lon($genbank_sequence);
}
process_locality($genbank_sequence);
break;
default:
break;
}
}
if (isset($genbank_sequence->source->latitude)) {
$hits->geometry->coordinates[] = array($genbank_sequence->source->longitude, $genbank_sequence->source->latitude);
}
$hits->sequences[] = $genbank_sequence;
}
}
return $hits;
}
