Treefam
Gene
Summary
Package variables
No package variables defined.
Included modules
Scalar::Util qw ( weaken )
Synopsis
use Treefam::DBConnection;
my $dbc = new Treefam::DBConnection ();
my $gene_handle = $dbc->get_GeneHandle();
my $gene = $gene_handle->get_by_id('ENSG00000087586');
my $symbol= $gene->symbol();
my @names = $gene->names();
my $family = $gene->family();
my $hmmer = $gene->hmmer_score();
my @orthologs = $gene->get_orthologs('Saccharomyces cerevisiae','Schizosaccharomyces pombe','Drosophila melanogaster','Caenorhabditis elegans');
Description
Representation of a Treefam gene.
Methods
Methods description
Arg: optional, Treefam gene ID
Description: Gets/sets Treefam ID
Returntype: string |
Arg: optional, chromosome as string
Description: Gets/sets chromosome that gene is on
Returntype: string |
Arg: optional, Treefam::Family object
Description: Gets/sets family the gene belongs to
Returntype: Treefam::Family object |
Arg: (optional) e-value cut-off
Description: Gets this gene's protein domains with e-value
below given cut-off, default is 1e-2.
Returntype: list of strings (PFAM domain IDs) |
Arg: optional, list of species (by latin name) from which to
get orthologs
Description: Gets this gene's orthologs
Returntype: list of Treefam::Gene objects |
Description: Gets this gene's paralogs. We currently only get
within-species paralogs
Returntype: list of Treefam::Gene objects |
Arg1: optional, Treefam::Family object or AC
Arg2: optional, hmmer evalue
Description: Gets/sets hmmer evalue for the gene in the given
family, defaults to the family the gene belongs to
Returntype: double |
Arg1: optional, Treefam::Family object or AC
Arg2: optional, hmmer score
Description: Gets hmmer score for the gene in the given family,
defaults to the family the gene belongs to
Returntype: double |
Arg: optional, names as one string
Description: Gets/sets the gene's names
Returntype: string |
Arg1: Treefam::DBConnection
Arg2: optional, gene ID
Description: Creates a new gene object.
Returntype: Treefam::Gene |
Arg: optional, integer position of gene (coding start) on
chromosome
Description: Gets/sets position of gene (coding start) on
chromosome. Returns 5'-most position in case
of multiple starts.
Returntype: integer |
Arg: optional, id of the gene's representative sequence
Description: Gets/sets the gene's representative sequence used
in the trees
Returntype: string |
Arg1: optional, type of name: latin or swcode
(5 letters species name used in Swissprot)
Arg2: optional, species name
Description: Gets/sets gene's species. Returns latin name
by default.
Returntype: string |
Arg: optional, gene symbol
Description: Gets/sets gene's symbol
Returntype: string |
Arg: optional, list of transcript IDs
Description: Gets/sets transcripts associated with gene
Returntype: list of transcript IDs |
Methods code
sub ID
{
my $self = shift;
$self->{'ID'} = shift if @_;
return $self->{'ID'};} | sub chromosome
{
my ($self,$chrom) = @_;
$self->{'chromosome'} = $chrom if $chrom;
if (!defined $self->{'chromosome'}) {
my $seqID = $self->sequence_id;
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $sth = $dbh->prepare("SELECT m.TARGET
FROM map m WHERE m.ID = ?");
$sth->execute($seqID);
($self->{'chromosome'})=$sth->fetchrow_array();
$sth->finish();
}
return $self->{'chromosome'};} | sub family
{
my $self = shift;
$self->{'family'} = shift if @_;
if (!defined $self->{'family'}) {
my $dbc = $self->{'DBConnection'};
my $famh = $dbc->get_FamilyHandle();
$self->{'family'} = $famh->get_by_gene($self);
}
return $self->{'family'};} | sub get_domains
{
my $self = shift;
my $cutoff = shift if @_;
if(!$cutoff) {
$cutoff = 1e-2;
}
if (!$self->{'domains'}) {
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $geneID = $self->sequence_id;
my %seen;
my $query = qq(SELECT DISTINCT PFAM_ID FROM pfam WHERE ID= ? AND EVALUE<$cutoff);
my $sth = $dbh->prepare($query);
$sth->execute($geneID);
while (my ($pfamid) = $sth->fetchrow_array()) {
push @{$self->{'domains'}},$pfamid if ($pfamid && !$seen{$pfamid}++);
}
}
@{$self->{'domains'}} = grep { $_ } @{$self->{'domains'}};
return @{$self->{'domains'}} if (defined($self->{'domains'}));} | sub get_orthologs
{
my $self = shift;
my @species = @_ if @_;
if (!defined $self->{'orthologs'}) {
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $geneID = $self->ID;
my $spec = $self->species();
my $genequery = qq(SELECT IDX FROM genes WHERE GID= ?);
my $sth0 = $dbh->prepare($genequery);
my $orthoquery1 = qq (SELECT DISTINCT g.GID, sp.taxname
FROM genes g, ortholog o, species sp
WHERE o.idx1 = ?
AND o.idx2 = g.idx
AND g.TAX_ID = sp.TAX_ID );
my $orthoquery2 = qq (SELECT DISTINCT g.GID, sp.taxname
FROM genes g, ortholog o, species sp
WHERE o.idx2 = ? AND o.idx1 = g.idx
AND g.TAX_ID = sp.TAX_ID );
my $sth1 = $dbh->prepare($orthoquery1);
my $sth2 = $dbh->prepare($orthoquery2);
$sth0->execute($geneID);
while (my ($idx) = $sth0->fetchrow_array()) {
foreach my $tmpsth($sth1,$sth2) {
$tmpsth->execute($idx);
while (my($orthologID,$species) = $tmpsth->fetchrow_array()) {
next if ($species eq $spec);
my $gh = $dbc->get_GeneHandle();
my $ortholog = $gh->get_by_id($orthologID);
if ($ortholog) {
push @{$self->{'orthologs'}},$ortholog;
$ortholog->species('latin',$species);
}
}
}
}
}
if (defined($self->{'orthologs'})) {
my %wanted;
if (@species) {
@wanted{@species} = (1) x @species;
@{$self->{'orthologs'}} = grep { $wanted{$_->species} } @{$self->{'orthologs'}};
}
return @{$self->{'orthologs'}};
}
else {
return ();
}} | sub get_paralogs
{
my $self = shift;
if (!defined $self->{'paralogs'}) {
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $geneID = $self->ID;
my $spec = $self->species();
my $genequery = qq(SELECT IDX FROM genes WHERE GID= ?);
my $sth0 = $dbh->prepare($genequery);
my $orthoquery1 = qq (SELECT DISTINCT g.GID, sp.taxname
FROM genes g, ortholog o, species sp
WHERE o.idx1 = ?
AND o.idx2 = g.idx
AND g.TAX_ID = sp.TAX_ID );
my $orthoquery2 = qq (SELECT DISTINCT g.GID, sp.taxname
FROM genes g, ortholog o, species sp
WHERE o.idx2 = ? AND o.idx1 = g.idx
AND g.TAX_ID = sp.TAX_ID );
my $sth1 = $dbh->prepare($orthoquery1);
my $sth2 = $dbh->prepare($orthoquery2);
$sth0->execute($geneID);
while(my ($idx) = $sth0->fetchrow_array()) {
foreach my $tmpsth($sth1,$sth2) {
$tmpsth->execute($idx);
while (my($paralogID,$species) = $tmpsth->fetchrow_array()) {
next if ($species ne $spec || $paralogID eq $geneID);
my $gh = $dbc->get_GeneHandle();
my $paralog = $gh->get_by_id($paralogID);
if ($paralog) {
push @{$self->{'paralogs'}},$paralog;
}
}
}
}
}
if (defined($self->{'paralogs'})) {
return @{$self->{'paralogs'}};
}
else {
return ();
}} | sub hmmer_evalue
{
my $self = shift;
my $family = shift if @_;
my $evalue = shift if @_;
if (!$family) {
$family = $self->family();
return undef unless $family;
}
my $familyID = ref($family)? $family->ID : $family;
if ($familyID && $evalue) {
$self->{'hmmer_evalue'}{$familyID} = $evalue;
}
if ($familyID && !defined $self->{'hmmer_evalue'}) {
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $geneID = $self->ID;
my $sth = $dbh->prepare("SELECT h.evalue
FROM hmmer h, genes g
WHERE g.gid = ?
AND g.id = h.id
AND h.ac = ?");
$sth->execute($geneID,$familyID);
($self->{'hmmer_evalue'}{$familyID}) = $sth->fetchrow_array();
$sth->finish();
}
return $self->{'hmmer_evalue'}{$familyID};} | sub hmmer_score
{
my $self = shift;
my $family = shift if @_;
my $score = shift if @_;
if (!$family) {
$family = $self->family();
return undef unless $family;
}
my $familyID = ref($family)? $family->ID : $family;
if ($familyID && $score) {
$self->{'hmmer_score'}{$familyID} = $score;
}
if ($familyID && !defined($self->{'hmmer_score'}{$familyID})) {
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $geneID = $self->ID;
my $sth = $dbh->prepare("SELECT h.score
FROM hmmer h, genes g
WHERE g.gid = ?
AND g.id = h.id
AND h.ac = ?");
$sth->execute($geneID,$familyID);
($self->{'hmmer_score'}{$familyID}) = $sth->fetchrow_array();
$sth->finish();
}
return $self->{'hmmer_score'}{$familyID} if $familyID;} | sub names
{
my ($self,$names) = @_;
$self->{'names'} = $names if $names;
if (!defined $self->{'names'}) {
my $geneID = $self->{'ID'};
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $sth = $dbh->prepare("SELECT DISTINCT g.desc
FROM genes g
WHERE g.gid= ?");
$sth->execute($geneID);
($self->{'names'})=$sth->fetchrow_array();
$sth->finish();
}
return $self->{'names'};} | sub new
{
my ($class,$dbc,$geneID) = @_;
my $self = {};
$self->{'DBConnection'} = $dbc;
weaken($self->{'DBConnection'});
my $dbh = $dbc->{'database_handle'};
$self->{'ID'} = $geneID;
bless ($self, $class);
return $self;} | sub position
{
my ($self,$pos) = @_;
$self->{'position'} = $pos if $pos;
if (!defined $self->{'position'}) {
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my @transcripts = $self->transcripts;
my $position;
my $sth = $dbh->prepare("SELECT m.C_START, m.C_STOP, m.STRAND
FROM map m, genes g
WHERE g.TID = ?
AND g.ID = m.ID");
foreach my $transcript(@transcripts) {
$sth->execute($transcript);
while ((my @array) = $sth->fetchrow_array) {
my $start = $array[0] + 1;
my $end = $array[1] + 1;
my $strand = $array[2];
if ($strand eq '-') {
($start,$end) = ($end,$start);
}
if (!(defined($position))) {
$position =$start;
}
else {
if ($strand eq '+') {
if ($start < $position) {
$position = $start;
}
}
elsif ($strand eq '-') {
if ($start > $position) {
$position = $start;
}
}
}
}
$sth->finish();
$self->{'position'} = $position;
}
}
return $self->{'position'};} | sub sequence_id
{
my $self = shift;
$self->{'sequence_id'} = shift if @_;
if (!defined $self->{'sequence_id'}) {
my $geneID = $self->{'ID'};
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $sth = $dbh->prepare("SELECT DISTINCT g.ID FROM genes g WHERE g.GID = ?");
$sth->execute($geneID);
($self->{'sequence_id'}) = $sth->fetchrow_array();
$sth->finish();
}
return $self->{'sequence_id'};} | sub species
{
my $self = shift;
my $type = shift if @_;
$type ||='latin';
$self->{'species'} = shift if @_;
if (!defined $self->{'species'} || (defined $self->{'species_name_type'} && defined $type && lc($type) ne lc($self->{'species_name_type'}))) {
my $geneID = $self->{'ID'};
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $sth;
if (defined($type) && lc($type) eq 'swcode') {
$sth = $dbh->prepare("SELECT DISTINCT s.swcode
FROM species s, genes g
WHERE g.GID = ?
AND g.tax_id=s.tax_id");
}
else {
$sth = $dbh->prepare("SELECT DISTINCT s.taxname
FROM species s, genes g
WHERE g.GID = ?
AND g.tax_id=s.tax_id");
}
$sth->execute($geneID);
($self->{'species'}) = $sth->fetchrow_array();
$sth->finish();
}
$self->{'species_name_type'} = defined($type) && lc($type) eq 'swcode' ? 'SWCODE' : 'LATIN';
return $self->{'species'};} | sub symbol
{
my $self = shift;
$self->{'symbol'} = shift if @_;
if (!defined $self->{'symbol'}) {
my $geneID = $self->{'ID'};
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $sth = $dbh->prepare("SELECT DISTINCT g.symbol FROM genes g WHERE g.GID = ?");
$sth->execute($geneID);
($self->{'symbol'}) = $sth->fetchrow_array();
$sth->finish();
}
return $self->{'symbol'};} | sub transcripts
{
my ($self,@transcripts) = @_;
push (@{$self->{'transcripts'}},@transcripts) if @transcripts;
if (!defined @{$self->{'transcripts'}}) {
my $geneID = $self->{'ID'};
my $dbc = $self->{'DBConnection'};
my $dbh = $dbc->{'database_handle'};
my $sth = $dbh->prepare("SELECT DISTINCT TID
FROM genes
WHERE gid= ?");
$sth->execute($geneID);
while (my ($transcript)=$sth->fetchrow_array()) {
push (@{$self->{'transcripts'}},$transcript);
}
}
return @{$self->{'transcripts'}};} | General documentation