feature Class Reference

#include <head.h>

List of all members.

Public Types

enum  Segrelation { none = 0, join, order }

Public Member Functions

void read (ifstream &ins, char ln[])
void write (ostream &ous) const
void clear ()
 feature ()
 feature (const feature &fe)
 ~feature ()
feature & operator= (const feature &fe)
int read (istream &ins, char nxt[])
int outSeg (ostream &ous) const
ostream & outRange (ostream &ous) const
ostream & outRange (ostream &ous, const char ed) const
ostream & endInfo (ostream &ous, const bool emark=true) const
void output (ostream &seq, ostream &sub, seqinfo &sinf) const
void dumpFeat (ostream &bd) const
void writeCDS (ostream &seq, ostream &sub, seqinfo &sinf) const
void writemRNA (ostream &seq, ostream &sub, seqinfo &sinf) const
void writeSource (ostream &seqout, ostream &subsout, seqinfo &sinf) const
void writeGene (ostream &seq, ostream &sub, seqinfo &sinf) const
void writeRNA (ostream &sout, ostream &subout, seqinfo &sinf) const
void writerpt (ostream &seq, ostream &sub, seqinfo &sinf) const
void writeExon (ostream &ous) const
void writeToken (ostream &seq, ostream &sub, seqinfo &sinf, const dseq &nt) const
void writeAllele (ostream &seq, ostream &sub, seqinfo &sinf) const
void wjoinseg (ostream &sub, seqinfo &sinf) const
void writeUTR (ostream &seq, ostream &sub, seqinfo &sinf) const
void wFeature (ostream &ous, const seqinfo &sinf) const
bool isAbnormal () const
bool complement () const
bool newfeat () const
bool is (const char *ff) const
bool isjoinseg () const
bool locIsJoin () const
const char * feat () const
int findex () const
bool findQualif (const char *qk, string &vv) const
string qvmap (const char *k) const
int qfind (const char *kq) const
 feature ()
 ~feature ()
bool next (string &ln, istream &ins)
 feature (const feature &feat)
feature & operator= (const feature &feat)
void writeAceProtein (ostream &ous, gbprtseq &prt)
void writeAceDNA (ostream &ous, ostream &sub, ostream &snp, gbdnaseq &seq) throw (featErr)
string getName () const
string getDbxrefString () const
string getAllQualifiers () const
void clear ()
int getBegin () const
int getEnd () const
int getNumSeg () const
const string & getOperation () const
Segrelation getSegop () const
ostream & writeRange (ostream &ous) const
ostream & outRange (ostream &ous) const
int outSeg (ostream &ous) const
void outSegMultiple (ostream &ous, const gbdnaseq &seq) const
string getLocationString () const
string getTaxid () const
bool hasDbxref () const
bool isComplement () const
int getBeginEndinfo () const
int getEndEndinfo () const
ostream & endInfo (ostream &ous) const
bool nostart () const
bool noend () const
bool hasQualifier (const string &key) const
string getQualifierValue (const string &qkey) const

Static Public Member Functions

static void readValidQualifier (const string &file)
static void loadSubseq (const string &file)
static void dumpSubseq (const string &file)
static void addSubseq (const string &seq)
static bool existSubseq (const string &seq)
static void loadGeneName (const string &file)
static void dumpGeneName (const string &file)
static string getGeneSymbol (const string &gene, string &allele)
static pair< string, string > getGeneSymbol (const string &gene)

Static Public Attributes

static const int QUAL = 5
static const int QUAL_VAL = 21
static const string QUAL_SPACE = string(QUAL_VAL, ' ')
static bool PRTOUT = false

Private Member Functions

void copy (const feature &lo)
int readloc (istream &ins, char ln[])
int qindex (const char qual[]) const
bool mktitle (string &tt) const
int subkey (string &sk) const
bool tokey (string &tk) const
void rptdump (ostream &ous, const seqinfo &sinf) const
void RNAdump (ostream &ous, const seqinfo &sinf) const
void subCDS (ostream &seq, ostream &sub, const string &cdsKey, const string &title, int tlnIdx, const seqinfo &sqinf) const
void nextQualifier (istream &ins, string &ln)
pair< string, string > geneallele () const
void outgeneline (ostream &ous, const gbdnaseq &seq, const string &tag) const
map< string, string >::iterator composeTitle ()
void insertQualifier (const string &key, string &value)
void writeSource (ostream &ous, ostream &sub, gbdnaseq &seq) throw (featErr)
void writemRNA (ostream &ous, ostream &sub, const gbdnaseq &seq)
void writeSegmRNA (ostream &sub, const gbdnaseq &seq)
void writeCDS (ostream &ous, ostream &sub, const gbdnaseq &seq)
void writeSegCDS (ostream &sub, const gbdnaseq &seq)
void subCDS (ostream &ous, ostream &sub, const gbdnaseq &seq, const string &cdsKey, const string &prtKey, const string &title) const
void writeGene (ostream &ous, ostream &sub, gbdnaseq &seq) const
void writeProtein (ostream &sub, const gbdnaseq &seq, const string &key, const string &title) const
void writeRNA (ostream &sout, ostream &subout, gbdnaseq &seq)
void RNAdump (ostream &ous, const gbdnaseq &seq) const
void writeFeature (ostream &ous, const gbdnaseq &seq) const throw (featErr)
bool writeFeatureOfWholeSeq (ostream &ous, const gbdnaseq &seq) const
void writeImmuno (ostream &ous, ostream &sub, const gbdnaseq &seq) const
void writeSNP (ostream &snp, gbdnaseq &seq) const
void writePrtSite (ostream &ous, gbprtseq &prt) const
void writePrtRegion (ostream &ous, gbprtseq &prt) const
void parseLoc () throw (featLocErr)

Static Private Member Functions

static string addGene (const string &gene, string &allele)
static pair< string, string > separateAllele (const string &gene)
static pair< string, string > makeGeneSymbol (const string &gene)
static pair< string, string > insertGeneSymbol (const string &gene, const string &symbol)
static pair< string, string > insertGeneAsSymbol (const string &gene)
static string geneOfSymbol (const string &sym)
static string cleanGene (const string &gene, const string wd)
static string nojunkGene (const string &gene)
static string nextGeneSymbol ()
static void goodGeneSymbol (const string &gene, string &symbol)
static bool isGoodGeneSymbol (const string &gene, const string &symbol)
static string acronymWithAllDigits (const string &str, const int n=1)
static string xHiddenGeneSymbol (const string &str, const string &sep)
static string shortIsGeneSymbol (const string &str, string::size_type idx, const int seplen)
static bool isRomanNumber (const string &str)
static bool isGreek (const string &str)

Private Attributes

string loc
string feat
int nq
qualif * qv
vector< string > dbxref
vector< string > qnv
int begin
int vb
int end
int ve
bool noLeft
bool noRight
string replseq
int * loc
int n
char * locstr
int l
int maxl
char curFeat [16]
string name
string locstr
vector< locseg * > locs
bool onone
bool complement
Segrelation segop
string locop
map< string, string > qualifiers

Static Private Attributes

static const char * ft [64]
static const char * qt [74]
static set< string > validqual = set<string>()
static map< string, int > subseq = map<string, int>()
static map< string, string > genen2s = map<string,string>()
static set< string > genesbl = set<string>()
static int genecnt = 0

Friends

const string orgtag (const seqinfo &sinf)
ostream & operator<< (ostream &ous, const feature &feat)

Member Enumeration Documentation

enum feature::Segrelation

Enumerator:
none 
join 
order 

Constructor & Destructor Documentation

feature::feature (  ) 

References curFeat.

feature::feature ( const feature &  fe  ) 

References copy().

feature::~feature (  ) 

References loc, locstr, and qv.

feature::feature (  )  [inline]

feature::~feature (  )  [inline]

References clear().

feature::feature ( const feature &  feat  ) 

Member Function Documentation

void feature::read ( ifstream &  ins,
char  ln[] 
)

References feat(), LINE, loc, and parerr::pick().

Referenced by main(), and protein::read().

void feature::write ( ostream &  ous  )  const

References feat(), and loc.

void feature::clear (  ) 

References feat(), and loc.

Referenced by gbseq::read(), protein::read(), and ~feature().

feature & feature::operator= ( const feature &  fe  ) 

References copy(), loc, locstr, and qv.

int feature::read ( istream &  ins,
char  nxt[] 
)

References dbxref, LINE, noLeft, noRight, nq, qnv, qv, qualif::qvalue, and readloc().

int feature::outSeg ( ostream &  ous  )  const

output Source_Exons num1 num2 will output the newline returns the sum_of_exon lengths

References begin, complement(), end, loc, and n.

Referenced by subCDS(), writeImmuno(), writemRNA(), and writeRNA().

ostream & feature::outRange ( ostream &  ous  )  const

output the range of this feature begin end, according to whether the sequence is complement or not. If complement will output end--begin will not output the newline

References begin, complement(), and end.

Referenced by outRange(), subCDS(), wFeature(), writeAllele(), writeCDS(), writeExon(), writeFeature(), writeFeatureOfWholeSeq(), writeGene(), writeImmuno(), writemRNA(), writeRNA(), writerpt(), writeSource(), and writeToken().

ostream& feature::outRange ( ostream &  ous,
const char  ed 
) const [inline]

add an termination mark ed, such as

References outRange().

ostream & feature::endInfo ( ostream &  ous,
const bool  emark = true 
) const

output start_not_found end_not_found to ous if emark = true,
is outputed at the end. Default emark true output seq end information to the ous if the sequence is not marked as missing start or end, nothing will be outputed to ous.

codon_start will be outputed if not 1; for CDS entries only

References complement(), noLeft, noRight, and qvmap().

Referenced by RNAdump(), subCDS(), wFeature(), writeCDS(), writeExon(), and writemRNA().

void feature::output ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

master feature output method can automatically select the right output function

References curFeat, is(), wFeature(), writeAllele(), writeCDS(), writeExon(), writeGene(), writemRNA(), writeRNA(), writerpt(), writeSource(), and writeUTR().

Referenced by main().

void feature::dumpFeat ( ostream &  bd  )  const

References curFeat, end, locstr, nq, qnv, and qv.

Referenced by main(), writeAllele(), and writeGene().

void feature::writeCDS ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

output CDS feature to sequence and subsequece files some change in ACEDB 4.9. Start_not_found 3 to indicate a frame start at 3, change CDS from 3 300 to 1 300

References begin, complement(), dbxref, end, endInfo(), seqinfo::getcntstr(), is(), l, locIsJoin(), mktitle(), seqinfo::name(), noLeft, noRight, seqinfo::org, outRange(), qfind(), qnv, qv, qualif::qvalue, qvmap(), seqinfo::sk, subCDS(), subkey(), and seqinfo::type.

Referenced by output(), and writeAceDNA().

void feature::writemRNA ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

References dbxref, endInfo(), findQualif(), seqinfo::getcntstr(), qualif::keyis(), mktitle(), seqinfo::name(), nq, orgtag, outRange(), outSeg(), featerr::pick(), qualif::qkey, qnv, qv, qualif::qvalue, seqinfo::sk, subkey(), and seqinfo::type.

Referenced by output(), and writeAceDNA().

void feature::writeSource ( ostream &  seqout,
ostream &  subsout,
seqinfo &  sinf 
) const

References begin, end, itoa(), qualif::keyis(), seqinfo::len, seqinfo::name(), nq, outRange(), qfind(), qualif::qkey, qnv, qv, qualif::qvalue, and seqinfo::sourCount.

Referenced by output(), and writeAceDNA().

void feature::writeGene ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

References dbxref, dumpFeat(), seqinfo::name(), nq, orgtag, outRange(), featerr::pick(), qfind(), qualif::qkey, qnv, qv, qualif::qvalue, split(), and seqinfo::type.

Referenced by output(), and writeAceDNA().

void feature::writeRNA ( ostream &  sout,
ostream &  subout,
seqinfo &  sinf 
) const

References begin, curFeat, end, itoa(), seqinfo::len, seqinfo::name(), outRange(), qfind(), qv, qualif::qvalue, seqinfo::rnaCount, RNAdump(), seqinfo::sk, and subkey().

Referenced by output(), and writeAceDNA().

void feature::writerpt ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

make a unique key for repeatsequence//////

make a title for repeat_region

References curFeat, seqinfo::getcntstr(), seqinfo::name(), outRange(), qfind(), qv, qualif::qvalue, seqinfo::sk, and subkey().

Referenced by output().

void feature::writeExon ( ostream &  ous  )  const

References curFeat, endInfo(), noLeft, noRight, nq, outRange(), qualif::qkey, qv, and qualif::qvalue.

Referenced by output().

void feature::writeToken ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf,
const dseq &  nt 
) const

References begin, curFeat, end, is(), itoa(), seqinfo::name(), nq, orgtag, outRange(), featerr::pick(), qualif::qkey, qnv, qv, qualif::qvalue, qvmap(), dseq::subseq(), seqinfo::tokCount, and tokey().

void feature::writeAllele ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

dealing with allele and variation feature may create an allele object if there are enough info

can make a title, or frequency creating an allele obj

References dumpFeat(), itoa(), mktitle(), seqinfo::name(), nq, outRange(), qfind(), qualif::qkey, qnv, qv, qualif::qvalue, qvmap(), replseq, seqinfo::tokCount, and tokey().

Referenced by output().

void feature::wjoinseg ( ostream &  sub,
seqinfo &  sinf 
) const

References seqinfo::cdsCount, curFeat, find(), is(), seqinfo::isseg(), itoa(), locstr, mktitle(), seqinfo::mrnaCount, seqinfo::name(), seqinfo::nameseg(), orgtag, qfind(), qualif::qkey, qv, qualif::qvalue, split(), and subkey().

Referenced by main().

void feature::writeUTR ( ostream &  seq,
ostream &  sub,
seqinfo &  sinf 
) const

most 5'UTR in the genomic annotation is useless, I will discard these

References locIsJoin(), and wFeature().

Referenced by output().

void feature::wFeature ( ostream &  ous,
const seqinfo &  sinf 
) const

All features other than source, CDS, mRNA, gene, RNA, repeat_region, satellite, exon, intron, allele, variation goes here, will be attached to EMBL_feature. ous is the main sequence output stream map to qualifer int int text EMBL_info

generic method for left over features

References begin, curFeat, dbxref, end, endInfo(), is(), qualif::keyis(), seqinfo::len, noLeft, noRight, nq, outRange(), gberr::pick(), qualif::qkey, qv, and qualif::qvalue.

Referenced by output(), and writeUTR().

bool feature::isAbnormal (  )  const

if location string contains num,letter gi, order, group, one-of, then it is considered abnormal

References locstr.

Referenced by readloc(), writeAceProtein(), writePrtRegion(), and writePrtSite().

bool feature::complement (  )  const

Referenced by endInfo(), getBeginEndinfo(), getEndEndinfo(), isComplement(), noend(), nostart(), outRange(), outSeg(), parseLoc(), writeCDS(), writemRNA(), and writeProtein().

bool feature::newfeat (  )  const [inline]

References findex().

Referenced by main().

bool feature::is ( const char *  ff  )  const [inline]

References curFeat.

Referenced by main(), output(), wFeature(), wjoinseg(), writeCDS(), and writeToken().

bool feature::isjoinseg (  )  const

location contains join of segment from different sequences

References locstr.

Referenced by main().

bool feature::locIsJoin (  )  const [inline]

References n.

Referenced by writeCDS(), and writeUTR().

const char* feature::feat (  )  const [inline]

References curFeat.

Referenced by clear(), read(), and write().

int feature::findex (  )  const

References curFeat, and ft.

Referenced by newfeat().

bool feature::findQualif ( const char *  qk,
string &  vv 
) const

put the qualifer value of pointed to by key qk to vv string

References qfind(), qv, and qualif::qvalue.

Referenced by mktitle(), subkey(), and writemRNA().

string feature::qvmap ( const char *  k  )  const

returns qvalue pointed to by qkey k. qkey is one of those in qualifier table. If not found, returns empty string.

References qfind(), qv, qualif::qvalue, and string().

Referenced by endInfo(), writeAllele(), writeCDS(), and writeToken().

int feature::qfind ( const char *  kq  )  const

look in the qv array for qkey, returns index of kq, returns -1 if not found

References nq, and qv.

Referenced by findQualif(), qvmap(), subCDS(), subkey(), tokey(), wjoinseg(), writeAllele(), writeCDS(), writeGene(), writeRNA(), writerpt(), and writeSource().

void feature::copy ( const feature &  lo  )  [private]

References begin, curFeat, dbxref, end, l, loc, locstr, maxl, n, noLeft, noRight, nq, qnv, qv, vb, and ve.

Referenced by feature(), and operator=().

int feature::readloc ( istream &  ins,
char  ln[] 
) [private]

before function call: ln must contain the first line of a feature ln is the input line, at the end of the function call, it will contain the next line following this feature this function also does the parsing of the location

return a error message. 0 for normal, 1 for parsing error

used by read()

joint segments

Bad stuff: 5'UTR join(2243..2291,2761) We now want to say this is ok. Represented with two pairs of integers

References append(), begin, curFeat, end, getNumber(), isAbnormal(), l, LINE, loc, locstr, maxl, n, noLeft, noRight, replseq, substr(), vb, and ve.

Referenced by read().

int feature::qindex ( const char  qual[]  )  const [private]

References qt.

bool feature::mktitle ( string &  tt  )  const [private]

References findQualif().

Referenced by wjoinseg(), writeAllele(), writeCDS(), and writemRNA().

int feature::subkey ( string &  sk  )  const [private]

References findQualif(), KEY_LIMIT, qfind(), qv, and qualif::qvalue.

Referenced by wjoinseg(), writeCDS(), writemRNA(), writeRNA(), writerpt(), and writeSource().

bool feature::tokey ( string &  tk  )  const [private]

makes a token key returns false if cannot make a keyy

References dbxref, qfind(), qv, and qualif::qvalue.

Referenced by writeAllele(), and writeToken().

void feature::rptdump ( ostream &  ous,
const seqinfo &  sinf 
) const [private]

References begin, end, qualif::keyis(), seqinfo::len, nq, orgtag, featerr::pick(), qnv, qv, and qualif::qvalue.

void feature::RNAdump ( ostream &  ous,
const seqinfo &  sinf 
) const [private]

References begin, curFeat, endInfo(), nq, orgtag, featerr::pick(), qnv, qv, and qualif::qvalue.

Referenced by writeRNA().

void feature::subCDS ( ostream &  seq,
ostream &  sub,
const string &  cdsKey,
const string &  title,
int  tlnIdx,
const seqinfo &  sqinf 
) const [private]

References endInfo(), nq, orgtag, outRange(), outSeg(), featerr::pick(), qfind(), qualif::qkey, qnv, qv, and qualif::qvalue.

Referenced by writeCDS().

bool feature::next ( string &  ln,
istream &  ins 
)

References locstr, name, nextQualifier(), parseLoc(), QUAL, QUAL_SPACE, QUAL_VAL, and trim().

Referenced by gbseq::read().

feature& feature::operator= ( const feature &  feat  ) 

void feature::writeAceProtein ( ostream &  ous,
gbprtseq &  prt 
)

this methods have dealt with: region, site, rest goest to Swiss_feature

References dbxref, getAllQualifiers(), getBegin(), getDbxrefString(), getEnd(), gbseq::getKey(), gbseq::getLength(), getLower(), getName(), gbseq::getOrgAcronym(), isAbnormal(), locs, name, qualifiers, writePrtRegion(), writePrtSite(), and writeRange().

void feature::writeAceDNA ( ostream &  ous,
ostream &  sub,
ostream &  snp,
gbdnaseq &  seq 
) throw (featErr)

References getNumSeg(), getQualifierValue(), name, featErr::print(), substr(), writeCDS(), writeFeature(), writeGene(), writeImmuno(), writemRNA(), writeRNA(), writeSNP(), and writeSource().

string feature::getName (  )  const [inline]

References name.

Referenced by writeAceProtein(), and writemRNA().

string feature::getDbxrefString (  )  const

References dbxref.

Referenced by writeAceProtein(), and writePrtRegion().

string feature::getAllQualifiers (  )  const

References qualifiers, and while().

Referenced by writeAceProtein(), writeFeatureOfWholeSeq(), and writePrtRegion().

void feature::clear (  ) 

mainly set optional members to defaults and clear up locs vector because it is a pointer array.

int feature::getBegin (  )  const [inline]

no complement information included For protein object this is enough

References locs.

Referenced by RNAdump(), writeAceProtein(), writeFeature(), writemRNA(), writeRange(), writeRNA(), writeSNP(), and writeSource().

int feature::getEnd (  )  const [inline]

References locs.

Referenced by writeAceProtein(), writeFeature(), writemRNA(), writeRange(), writeRNA(), writeSNP(), and writeSource().

int feature::getNumSeg (  )  const [inline]

References locs.

Referenced by writeAceDNA(), and writePrtRegion().

const string& feature::getOperation (  )  const [inline]

References locop.

Referenced by writeFeature().

Segrelation feature::getSegop (  )  const [inline]

References segop.

Referenced by writeCDS().

ostream& feature::writeRange ( ostream &  ous  )  const [inline]

References getBegin(), and getEnd().

Referenced by writeAceProtein(), and writeCDS().

ostream& feature::outRange ( ostream &  ous  )  const

int feature::outSeg ( ostream &  ous  )  const

output the subsequence ranges for CDS and mRNA only, Format: one or more {Source_Exon begin end}

void feature::outSegMultiple ( ostream &  ous,
const gbdnaseq &  seq 
) const

output coding region (exons) from multiple sequences

References gbseq::getKey(), and locs.

Referenced by writeSegCDS(), and writeSegmRNA().

string feature::getLocationString (  )  const [inline]

Return the GenBank style location exactly as read from input.

References locstr.

Referenced by writeFeature().

string feature::getTaxid (  )  const

try to extract taxid from the first source field May be a problem is mosaic sequences. This methods is only defined for feature Source

References dbxref, name, and substr().

bool feature::hasDbxref (  )  const [inline]

References dbxref.

bool feature::isComplement (  )  const [inline]

References complement().

int feature::getBeginEndinfo (  )  const

has complement considerred

References complement(), and locs.

Referenced by writemRNA().

int feature::getEndEndinfo (  )  const

References complement(), and locs.

Referenced by writemRNA().

ostream & feature::endInfo ( ostream &  ous  )  const

References complement(), and locs.

bool feature::nostart (  )  const

References complement(), and locs.

Referenced by RNAdump(), subCDS(), writeCDS(), and writeProtein().

bool feature::noend (  )  const

References complement(), and locs.

Referenced by RNAdump(), subCDS(), writeCDS(), and writeProtein().

bool feature::hasQualifier ( const string &  key  )  const [inline]

References qualifiers.

Referenced by subCDS(), writeCDS(), writeFeatureOfWholeSeq(), and writeSegCDS().

string feature::getQualifierValue ( const string &  qkey  )  const

References qualifiers, and string().

Referenced by geneallele(), outgeneline(), writeAceDNA(), writeGene(), and writeSegCDS().

void feature::readValidQualifier ( const string &  file  )  [static]

References ifstream(), string(), and validqual.

void feature::loadSubseq ( const string &  file  )  [static]

Load subsequence keys from a file containing all relevant subsequence keys, so that the parser knows what subsequences have been parsed. Keeps track of globally named subsequences.

References ifstream(), and subseq.

Referenced by main().

void feature::dumpSubseq ( const string &  file  )  [static]

References subseq.

Referenced by main().

static void feature::addSubseq ( const string &  seq  )  [inline, static]

References subseq.

Referenced by writeCDS(), and writemRNA().

static bool feature::existSubseq ( const string &  seq  )  [inline, static]

References subseq.

void feature::loadGeneName ( const string &  file  )  [static]

functions working with gene names

GeneName file use tab delimited file geneName

References genen2s, genesbl, ifstream(), and split().

Referenced by main().

void feature::dumpGeneName ( const string &  file  )  [static]

References genen2s.

Referenced by main().

string feature::getGeneSymbol ( const string &  gene,
string &  allele 
) [static]

will check whether gene is a symbol or not if not will insert it into gensbl and add to genen2s. This is the interface that will be used by the public.

References genen2s, makeGeneSymbol(), and wc().

Referenced by geneallele(), and outgeneline().

pair< string, string > feature::getGeneSymbol ( const string &  gene  )  [static]

References genen2s, makeGeneSymbol(), and wc().

static string feature::addGene ( const string &  gene,
string &  allele 
) [static, private]

If getGeneSymbol() gets nothing it calles this methods. Add a new gene into genen2s map<string,string>. Returns the symbol for gene. Some gene are actually allele and this will be separated. Looks like * separates the gene name from allele gene*allele

pair< string, string > feature::separateAllele ( const string &  gene  )  [static, private]

only if you are sure this is a legal Gene*symbol structure you can call this function.

References wc().

Referenced by insertGeneSymbol().

pair< string, string > feature::makeGeneSymbol ( const string &  gene  )  [static, private]

References acronymWithAllDigits(), delall(), dissect(), firstword(), genen2s, genesbl, goodGeneSymbol(), insertGeneSymbol(), isGreek(), isnumber(), isRomanNumber(), isupper(), lastword(), length, nextGeneSymbol(), nojunkGene(), split(), toupper(), tr(), wc(), and xHiddenGeneSymbol().

Referenced by getGeneSymbol().

pair< string, string > feature::insertGeneSymbol ( const string &  gene,
const string &  symbol 
) [static, private]

References genen2s, genesbl, separateAllele(), and tr().

Referenced by insertGeneAsSymbol(), and makeGeneSymbol().

pair< string, string > feature::insertGeneAsSymbol ( const string &  gene  )  [static, private]

References insertGeneSymbol().

string feature::geneOfSymbol ( const string &  sym  )  [static, private]

actually more than one gene may point to the same symbol this function returns the first one found

References genen2s, and string().

Referenced by isGoodGeneSymbol().

string feature::cleanGene ( const string &  gene,
const string  wd 
) [static, private]

References singleSpace().

Referenced by nojunkGene().

string feature::nojunkGene ( const string &  gene  )  [static, private]

References cleanGene(), isGreek(), L, lastword(), and singleSpace().

Referenced by goodGeneSymbol(), isGoodGeneSymbol(), and makeGeneSymbol().

string feature::nextGeneSymbol (  )  [static, private]

References genecnt, genesbl, and itos().

Referenced by makeGeneSymbol().

void feature::goodGeneSymbol ( const string &  gene,
string &  symbol 
) [static, private]

References acronymWithAllDigits(), genesbl, isGoodGeneSymbol(), itos(), l, and nojunkGene().

Referenced by makeGeneSymbol().

bool feature::isGoodGeneSymbol ( const string &  gene,
const string &  symbol 
) [static, private]

References cmp_nocase(), firstword(), geneOfSymbol(), genesbl, lastword(), and nojunkGene().

Referenced by goodGeneSymbol().

string feature::acronymWithAllDigits ( const string &  str,
const int  n = 1 
) [static, private]

References dissect(), isGreek(), isnumber(), isRomanNumber(), isupper(), and length.

Referenced by goodGeneSymbol(), and makeGeneSymbol().

string feature::xHiddenGeneSymbol ( const string &  str,
const string &  sep 
) [static, private]

References shortIsGeneSymbol().

Referenced by makeGeneSymbol().

string feature::shortIsGeneSymbol ( const string &  str,
string::size_type  idx,
const int  seplen 
) [static, private]

References acronym(), isupper(), strToupper(), and wc().

Referenced by xHiddenGeneSymbol().

bool feature::isRomanNumber ( const string &  str  )  [static, private]

Referenced by acronymWithAllDigits(), and makeGeneSymbol().

bool feature::isGreek ( const string &  str  )  [static, private]

Referenced by acronymWithAllDigits(), makeGeneSymbol(), and nojunkGene().

void feature::nextQualifier ( istream &  ins,
string &  ln 
) [private]

References insertQualifier(), QUAL_SPACE, QUAL_VAL, qualifiers, and value.

Referenced by next().

pair< string, string > feature::geneallele (  )  const [private]

used up gene and allele, or locus_tag qualifier if they exist otherwise return a pair of empty strings.

References getGeneSymbol(), getQualifierValue(), name, and qualifiers.

Referenced by writeGene().

void feature::outgeneline ( ostream &  ous,
const gbdnaseq &  seq,
const string &  tag 
) const [private]

for output the gene line, tag can be: geneof_mRNA, geneof_CDS, geneof_model, geneof_transcript or geneof_protein

References getGeneSymbol(), gbseq::getOrgAcronym(), getQualifierValue(), and qualifiers.

Referenced by RNAdump(), subCDS(), writeCDS(), writeFeatureOfWholeSeq(), writeImmuno(), writemRNA(), writeProtein(), writeRNA(), writeSegCDS(), and writeSegmRNA().

map< string, string >::iterator feature::composeTitle (  )  [private]

References l, length, maxl, and qualifiers.

Referenced by writeCDS(), writemRNA(), writeSegCDS(), and writeSegmRNA().

void feature::insertQualifier ( const string &  key,
string &  value 
) [private]

only needs to check the double quote at the end in case someone forgot to add the beginning quote if (value[0] == '"' && value[value.length()-1] == '"') value = value.substr(1, value.length()-2); else if (value[value.length()-1] == '"') value = value.substr(0, value.length()-1); else if (value[0] == '"') value = value.substr(1);

References dbxref, qualifiers, and tr().

Referenced by nextQualifier().

void feature::writeSource ( ostream &  ous,
ostream &  sub,
gbdnaseq &  seq 
) throw (featErr) [private]

References getBegin(), getEnd(), getLower(), itos(), length, outRange(), qualifiers, split(), and subkey().

void feature::writemRNA ( ostream &  ous,
ostream &  sub,
const gbdnaseq &  seq 
) [private]

when parsing Genomic DNA and mRNAs sometime times redundancy causes trouble. In the human genome annotation, the genome annotation may have a mRNA annotated on the genomic DNA. At the same time there is a mRNA dump that some times may have more than one segment. The polycy of this program make another subsequence as CDS and links to the protein sequence. There should be a way to link without the subsequence

References addSubseq(), complement(), composeTitle(), dbxref, locseg::getBegin(), getBegin(), getBeginEndinfo(), locseg::getEnd(), getEnd(), getEndEndinfo(), locseg::getFuzzyBegin(), locseg::getFuzzyEnd(), gbseq::getKey(), gbseq::getLength(), getName(), gbseq::getType(), itos(), join, locs, gbdnaseq::nextsub(), onone, outgeneline(), outRange(), outSeg(), qualifiers, segop, subseq, and writeSegmRNA().

void feature::writeSegmRNA ( ostream &  sub,
const gbdnaseq &  seq 
) [private]

References composeTitle(), gbseq::getKey(), itos(), gbdnaseq::nextsub(), outgeneline(), outSegMultiple(), and qualifiers.

Referenced by writemRNA().

void feature::writeCDS ( ostream &  ous,
ostream &  sub,
const gbdnaseq &  seq 
) [private]

If the CDS is from a mRNA and have more than one segments, it is more likely specifying a translational exception. In this case the distance between segments is usually < 3 and usually have two segments. I will create a subsequence for this mRNA and should not name it according to the corresponding protein.

References addSubseq(), complement(), composeTitle(), gbseq::getKey(), gbseq::getLocusName(), gbseq::getMolType(), getSegop(), hasQualifier(), itos(), join, gbdnaseq::nextsub(), noend(), nostart(), onone, outgeneline(), outRange(), PRTOUT, qualifiers, segop, subCDS(), subkey(), subseq, writeProtein(), writeRange(), and writeSegCDS().

void feature::writeSegCDS ( ostream &  sub,
const gbdnaseq &  seq 
) [private]

References composeTitle(), gbseq::getKey(), getQualifierValue(), hasQualifier(), itos(), gbdnaseq::nextsub(), outgeneline(), outSegMultiple(), PRTOUT, qualifiers, and writeProtein().

Referenced by writeCDS().

void feature::subCDS ( ostream &  ous,
ostream &  sub,
const gbdnaseq &  seq,
const string &  cdsKey,
const string &  prtKey,
const string &  title 
) const [private]

A helper function to write CDS object Add one more parameter to handle where CDS and protein have different names. This is useful when dealing with translation frame-shifts, where the CDS for splicing is different from CDS for translation, but these two different CDS object points to the same protein.

References dbxref, gbseq::getKey(), gbseq::getOrgAcronym(), hasQualifier(), noend(), nostart(), outgeneline(), outRange(), outSeg(), qualifiers, and split().

void feature::writeGene ( ostream &  ous,
ostream &  sub,
gbdnaseq &  seq 
) const [private]

References dbxref, geneallele(), gbseq::getKey(), gbseq::getLocusName(), gbseq::getMolType(), gbseq::getOrgAcronym(), getQualifierValue(), gbseq::getType(), locs, name, onone, outRange(), qualifiers, and split().

void feature::writeProtein ( ostream &  sub,
const gbdnaseq &  seq,
const string &  key,
const string &  title 
) const [private]

helper function

use the CDS feature to construct a protein object May or may not have Peptide Object associated with it. Borrow title from CDS object

References complement(), dbxref, gbseq::getKey(), gbseq::getOrgAcronym(), gbseq::getOrganism(), noend(), nostart(), outgeneline(), qualifiers, split(), and writeSequence().

Referenced by writeCDS(), and writeSegCDS().

void feature::writeRNA ( ostream &  sout,
ostream &  subout,
gbdnaseq &  seq 
) [private]

References dbxref, getBegin(), getEnd(), gbseq::getKey(), gbseq::getLength(), gbseq::getType(), gbseq::hasFeature(), itos(), name, gbdnaseq::nextsub(), outgeneline(), outRange(), outSeg(), qualifiers, and RNAdump().

void feature::RNAdump ( ostream &  ous,
const gbdnaseq &  seq 
) const [private]

References dbxref, getBegin(), name, noend(), nostart(), outgeneline(), qualifiers, split(), and substr().

void feature::writeFeature ( ostream &  ous,
const gbdnaseq &  seq 
) const throw (featErr) [private]

References dbxref, getBegin(), getEnd(), getLocationString(), getOperation(), locs, name, outRange(), qualifiers, tr(), and writeFeatureOfWholeSeq().

Referenced by writeAceDNA().

bool feature::writeFeatureOfWholeSeq ( ostream &  ous,
const gbdnaseq &  seq 
) const [private]

References getAllQualifiers(), gbseq::getKey(), hasQualifier(), name, outgeneline(), outRange(), and qualifiers.

Referenced by writeFeature().

void feature::writeImmuno ( ostream &  ous,
ostream &  sub,
const gbdnaseq &  seq 
) const [private]

References gbseq::getKey(), itos(), name, gbdnaseq::nextsub(), outgeneline(), outRange(), outSeg(), and qualifiers.

Referenced by writeAceDNA().

void feature::writeSNP ( ostream &  snp,
gbdnaseq &  seq 
) const [private]

References dbxref, getBegin(), getEnd(), gbseq::getKey(), and qualifiers.

Referenced by writeAceDNA().

void feature::writePrtSite ( ostream &  ous,
gbprtseq &  prt 
) const [private]

References dbxref, isAbnormal(), locs, and qualifiers.

Referenced by writeAceProtein().

void feature::writePrtRegion ( ostream &  ous,
gbprtseq &  prt 
) const [private]

References dbxref, getAllQualifiers(), getDbxrefString(), getNumSeg(), isAbnormal(), locs, and qualifiers.

Referenced by writeAceProtein().

void feature::parseLoc (  )  throw (featLocErr) [private]

for normal locations on one sequence

References complement(), join, locs, locstr, onone, order, segop, and split().

Referenced by next().

Friends And Related Function Documentation

const string orgtag ( const seqinfo &  sinf  )  [friend]

Referenced by RNAdump(), rptdump(), subCDS(), wjoinseg(), writeGene(), writemRNA(), and writeToken().

ostream& operator<< ( ostream &  ous,
const feature &  feat 
) [friend]

ous << feat.loc;

Member Data Documentation

string feature::loc [private]

Referenced by clear(), copy(), operator=(), outSeg(), read(), readloc(), write(), and ~feature().

string feature::feat [private]

int feature::nq [private]

Referenced by copy(), dumpFeat(), qfind(), read(), RNAdump(), rptdump(), subCDS(), wFeature(), writeAllele(), writeExon(), writeGene(), writemRNA(), writeSource(), and writeToken().

qualif* feature::qv [private]

Referenced by copy(), dumpFeat(), findQualif(), operator=(), qfind(), qvmap(), read(), RNAdump(), rptdump(), subCDS(), subkey(), tokey(), wFeature(), wjoinseg(), writeAllele(), writeCDS(), writeExon(), writeGene(), writemRNA(), writeRNA(), writerpt(), writeSource(), writeToken(), and ~feature().

vector< string > feature::dbxref [private]

Referenced by copy(), getDbxrefString(), getTaxid(), hasDbxref(), insertQualifier(), operator<<(), read(), RNAdump(), subCDS(), tokey(), wFeature(), writeAceProtein(), writeCDS(), writeFeature(), writeGene(), writemRNA(), writeProtein(), writePrtRegion(), writePrtSite(), writeRNA(), and writeSNP().

vector<string> feature::qnv [private]

Referenced by copy(), dumpFeat(), read(), RNAdump(), rptdump(), subCDS(), writeAllele(), writeCDS(), writeGene(), writemRNA(), writeSource(), and writeToken().

int feature::begin [private]

Referenced by copy(), outRange(), outSeg(), readloc(), RNAdump(), rptdump(), wFeature(), writeCDS(), writeRNA(), writeSource(), and writeToken().

int feature::vb [private]

Referenced by copy(), and readloc().

int feature::end [private]

Referenced by copy(), dumpFeat(), outRange(), outSeg(), readloc(), rptdump(), wFeature(), writeCDS(), writeRNA(), writeSource(), and writeToken().

int feature::ve [private]

Referenced by copy(), and readloc().

bool feature::noLeft [private]

Referenced by copy(), endInfo(), read(), readloc(), wFeature(), writeCDS(), and writeExon().

bool feature::noRight [private]

Referenced by copy(), endInfo(), read(), readloc(), wFeature(), writeCDS(), and writeExon().

string feature::replseq [private]

Referenced by readloc(), and writeAllele().

int* feature::loc [private]

int feature::n [private]

Referenced by copy(), locIsJoin(), outSeg(), and readloc().

char* feature::locstr [private]

Referenced by copy(), dumpFeat(), getLocationString(), isAbnormal(), isjoinseg(), next(), operator=(), parseLoc(), readloc(), wjoinseg(), and ~feature().

int feature::l [private]

Referenced by composeTitle(), copy(), goodGeneSymbol(), readloc(), and writeCDS().

int feature::maxl [private]

Referenced by composeTitle(), copy(), and readloc().

char feature::curFeat[16] [private]

Referenced by copy(), dumpFeat(), feat(), feature(), findex(), is(), output(), readloc(), RNAdump(), wFeature(), wjoinseg(), writeExon(), writeRNA(), writerpt(), and writeToken().

const char * feature::ft [static, private]

Initial value:

 { "3'UTR", "3'clip",   "5'UTR", 
"5'clip", "CAAT_signal", "CDS",   "C_region", "D-loop",
"D_segment",   "GC_signal",  "J_region", "J_segment", "LTR",
"N_region", "RBS",   "STS",   "S_region",
"TATA_signal", "V_region", "V_segment",   "allele",
"attenuator",  "conflict", "enhancer", "exon",
"gene",  "iDNA",  "intron",   "mRNA",
"mat_peptide", "misc_RNA", "misc_binding",   "misc_difference",
"misc_feature",   "misc_recomb", "misc_signal", "misc_structure",
"modified_base",  "mutation", "old_sequence",   "polyA_signal",
"polyA_site",  "precursor_RNA",  "prim_transcript",   "primer",
"primer_bind", "promoter", "protein_bind",   "rRNA",
"rep_origin",  "repeat_region",  "repeat_unit", "satellite",
"scRNA", "sig_peptide", "snRNA", "source",
"stem_loop",   "tRNA",  "terminator",  "transit_peptide",
"transposon",  "unsure",   "variation" }

Referenced by findex().

const char * feature::qt [static, private]

Initial value:

 {"EC_number",   
"PCR_conditions", "allele",   "anticodon",
"bound_moiety",   "cell_line",   "cell_type",   "chloroplast",
"chromoplast", "chromosome",  "citation", "clone",
"clone_lib",   "codon", "codon_start", "cons_splice",
"cultivar", "cyanelle", "db_xref",  "dev_stage",
"direction",   "evidence", "exception",   "focus",
"frequency",   "function", "gdb_xref", "gene",
"germline", "haplotype",   "insertion_seq",  "isolate",
"kinetoplast", "lab_host", "label", "macronuclear",
"map",   "mitochondrion",  "mod_base", "note",
"number",   "organism", "partial",  "phenotype",
"plasmid",  "pop_variant", "product",  "protein_id",
"proviral", "pseudo",   "rearranged",  "replace",
"rpt_family",  "rpt_type", "rpt_unit", "sequenced_mol",
"serotype", "sex",   "specific_host",  "specimen_voucher",
"standard_name",  "strain",   "sub_clone",   "sub_species",
"sub_strain",  "tissue_lib",  "tissue_type", "transl_except",
"transl_table",   "translation", "transposon",  "usedin",
"variety",  "virion"}

Referenced by qindex().

const int feature::QUAL = 5 [static]

Referenced by next().

const int feature::QUAL_VAL = 21 [static]

Referenced by next(), and nextQualifier().

const string feature::QUAL_SPACE = string(QUAL_VAL, ' ') [static]

Referenced by next(), and nextQualifier().

PRTOUT is a boolean variable P set feature::PRTOUT = false [static]

Referenced by main(), writeCDS(), and writeSegCDS().

string feature::name [private]

Referenced by geneallele(), getName(), getTaxid(), next(), operator<<(), RNAdump(), writeAceDNA(), writeAceProtein(), writeFeature(), writeFeatureOfWholeSeq(), writeGene(), writeImmuno(), and writeRNA().

string feature::locstr [private]

vector<locseg*> feature::locs [private]

To provide polymorphism, I must use pointers for the location. the Derived class namedlocseg is rarely used.

Referenced by endInfo(), getBegin(), getBeginEndinfo(), getEnd(), getEndEndinfo(), getNumSeg(), noend(), nostart(), operator<<(), outSegMultiple(), parseLoc(), writeAceProtein(), writeFeature(), writeGene(), writemRNA(), writePrtRegion(), and writePrtSite().

bool feature::onone [private]

Referenced by parseLoc(), writeCDS(), writeGene(), and writemRNA().

bool feature::complement [private]

Segrelation feature::segop [private]

Referenced by getSegop(), parseLoc(), writeCDS(), and writemRNA().

string feature::locop [private]

Referenced by getOperation(), and operator<<().

map<string, string> feature::qualifiers [private]

Referenced by composeTitle(), geneallele(), getAllQualifiers(), getQualifierValue(), hasQualifier(), insertQualifier(), nextQualifier(), operator<<(), outgeneline(), RNAdump(), subCDS(), writeAceProtein(), writeCDS(), writeFeature(), writeFeatureOfWholeSeq(), writeGene(), writeImmuno(), writemRNA(), writeProtein(), writePrtRegion(), writePrtSite(), writeRNA(), writeSegCDS(), writeSegmRNA(), writeSNP(), and writeSource().

set< string > feature::validqual = set<string>() [static, private]

Referenced by readValidQualifier().

map< string, int > feature::subseq = map<string, int>() [static, private]

Referenced by addSubseq(), dumpSubseq(), existSubseq(), loadSubseq(), writeCDS(), and writemRNA().

map< string, string > feature::genen2s = map<string,string>() [static, private]

Gene Name to acronym mapping for long gene names if first is already a gene symbol then second is "" if gene have more than one word, use the first letter of each word to make up the gene symbol.

Referenced by dumpGeneName(), geneOfSymbol(), getGeneSymbol(), insertGeneSymbol(), loadGeneName(), and makeGeneSymbol().

set< string > feature::genesbl = set<string>() [static, private]

Referenced by goodGeneSymbol(), insertGeneSymbol(), isGoodGeneSymbol(), loadGeneName(), makeGeneSymbol(), and nextGeneSymbol().

int feature::genecnt = 0 [static, private]

Referenced by nextGeneSymbol().

The documentation for this class was generated from the following files:
Generated on Wed Aug 10 11:57:10 2011 for Softwares from Orpara by  doxygen 1.5.6