doublestop.cpp File Reference

#include <iostream>
#include <mysql++.h>
#include "dbinfo.h"
#include <cstring>
#include <string>
#include <set>
#include <map>
#include <vector>
#include "bioseq.h"
#include <cmath>
#include <boost/math/distributions.hpp>
#include <RNAModel.h>

Classes
class	Progparam
class	Conpos
class	Conseq
Functions
bool	loadGenomic (map< string, string > &gstore, Connection &conn, const string &host, const string &database)
bool	checkstop (const string &host, const string &dbname, const MysqlDBInfo &mydb, ostream &ous, ostream &txou, const Progparam &par)
bool	isStop (const string &codon)
int	stops (const string &seq)
bool	hasTable (Connection &conn, const string &tab)
void	readconf (const string &file, map< string, string > &dbtab, set< string > &ignore, map< string, int > &maxintron)
double	distanceBaseFreq (double bf1[4], double bf2[4])
void	checkBadStopIndex (const string &host, const string &dbname, const MysqlDBInfo &mydb, ostream &ous, ostream &osl, map< string, int > &maxintron, const Progparam &par)
void	analyzeStop (const string &host, const vector< string > &dbs, const MysqlDBInfo &mydb, const Progparam &par, bool appendres=false)
void	fixStopIndex (const string &host, vector< string > &dbs, const MysqlDBInfo &mydb, map< string, int > &maxintron, const Progparam &par)
void	refineAllmodels (Connection &conn, const string &fixedfile, const string &badfile, const string &refinedallmod)
void	getCandidateDatabase (vector< string > &dbnames, Connection &conn)
void	discardIgnored (vector< string > &dbnames, const set< string > &ignored)
void	getDbsFromFile (vector< string > &dbnames, const string &dblf)
void	usage ()
int	main (int argc, char *argv[])
ostream &	operator<< (ostream &ous, const Conpos &pos)
ostream &	operator<< (ostream &ous, const Conseq &cons)
void	countbase (int bs[4], const string &seq)
double	expectedStop (double bf[4], int n)
void	stoplocation (const string &seq, vector< int > &sloc)
int	count2freq (int bc[4], double bf[4])
template<class T>
ostream &	writeBase (T bf[4], ostream &ous, const char *sep="\t")
bool	loadGenomicFromScaffoldSeq (map< string, string > &gstore, Connection &conn)
ostream &	writeRow (Row row, ostream &ous)
ostream &	writeResult (ostream &ous, const Row &row, const JGIModel &mod)

---

Function Documentation

void analyzeStop	(	const string &	host,
		const vector< string > &	dbs,
		const MysqlDBInfo &	mydb,
		const Progparam &	par,
		bool	appendres = false
	)

References checkstop().

Referenced by main().

void checkBadStopIndex	(	const string &	host,
		const string &	dbname,
		const MysqlDBInfo &	mydb,
		ostream &	ous,
		ostream &	osl,
		map< string, int > &	maxintron,
		const Progparam &	par
	)

Parameters:

	ous.	For ouput of all databases.
	osl.	is the log steam for debug purpose. fix stop, left or right use all models This methods calls buildAllModels to build allmodels table.

It generates dbname + "_stopfixed.tab" and dbname + "_badmodels.tab"

References at, mRNAModel::complete(), Range::direction(), MysqlDBInfo::getPassword(), MysqlDBInfo::getUser(), mRNAModel::growCDS3Prime(), mRNAModel::hasStop(), itos(), mRNAModel::jgiModelCol, mRNAModel::jgiTranscriptCol, length, loadGenomic(), Progparam::nreptab, mRNAModel::proteinLength(), mRNAModel::proteinLengthNoTail(), mRNAModel::proteinSequence(), refineAllmodels(), Progparam::replaceallmod, string(), mRNAModel::trimCDSStop(), mRNAModel::trimCDSTail(), Badinput::what(), PointOutChain::what(), InvalidModel::what(), InvalidJGIModel::what(), and writeResult().

Referenced by fixStopIndex().

bool checkstop	(	const string &	host,
		const string &	dbname,
		const MysqlDBInfo &	mydb,
		ostream &	ous,
		ostream &	txou,
		const Progparam &	par
	)

Analysze the stop situation for one database.

Parameters:

ous

contains the summary reuslt for the stop region

Returns:

true if process successful. False if failed for some reason.

References mRNAModel::CDSLength(), mRNAModel::CDSSequence(), count2freq(), countbase(), Range::direction(), distanceBaseFreq(), expectedStop(), mRNAModel::genomicCDSEnd(), RNAModel::getOid(), MysqlDBInfo::getPassword(), MysqlDBInfo::getUser(), hasTable(), L, loadGenomic(), Progparam::nreptab, reverseComplement(), stoplocation(), string(), and writeBase().

Referenced by analyzeStop().

int count2freq	(	int	bc[4],
		double	bf[4]
	)

Referenced by checkstop().

void countbase	(	int	bs[4],
		const string &	seq
	)

References toupper().

Referenced by checkstop().

void discardIgnored	(	vector< string > &	dbnames,
		const set< string > &	ignored
	)

References good.

Referenced by main().

double distanceBaseFreq	(	double	bf1[4],
		double	bf2[4]
	)

Eucleadian distance betwee two base coverage profiles

Referenced by checkstop().

double expectedStop	(	double	bf[4],
		int	n
	)

since stop codons are not independent in different frames, we need to do a small correction even for random sequences If frame 1 has a stop, then the other two frames will not be a stop codon for sure. So the absolute frequence of stop codon given any random codon is the sum of the three codons. In a long sequence this needs to be corrected by the following equation: Pcorrected = p-(4np^2)/(3n-2)

Parameters:

	bf	base frequence of A,C,G, and T
	n	number of codons of sequence. seqlen/3.

Referenced by checkstop().

void fixStopIndex	(	const string &	host,
		vector< string > &	dbs,
		const MysqlDBInfo &	mydb,
		map< string, int > &	maxintron,
		const Progparam &	par
	)

this is the top level method, it can iterate through multiple databases. It uses fixStopIndex to work on each individual database.

References checkBadStopIndex().

Referenced by main().

void getCandidateDatabase	(	vector< string > &	dbnames,
		Connection &	conn
	)

References hasTable().

Referenced by main().

void getDbsFromFile	(	vector< string > &	dbnames,
		const string &	dblf
	)

References ifstream().

Referenced by main().

bool hasTable	(	Connection &	conn,
		const string &	tab
	)

Referenced by checkstop(), getCandidateDatabase(), and loadGenomicFromScaffoldSeq().

bool isStop

(

const string &

codon

)

Referenced by stoplocation(), and stops().

bool loadGenomic	(	map< string, string > &	gstore,
		Connection &	conn,
		const string &	host,
		const string &	database
	)

References length, loadGenomicFromScaffoldSeq(), and string().

Referenced by checkBadStopIndex(), and checkstop().

bool loadGenomicFromScaffoldSeq	(	map< string, string > &	gstore,
		Connection &	conn
	)

References hasTable(), and string().

Referenced by loadGenomic().

int main	(	int	argc,
		char *	argv[]
	)

References analyzeStop(), discardIgnored(), Progparam::fixstop, fixStopIndex(), MysqlDBInfo::getAuthenInfo(), getCandidateDatabase(), getDbsFromFile(), MysqlDBInfo::getPassword(), MysqlDBInfo::getUser(), Progparam::nreptab, readconf(), Progparam::replaceallmod, mRNAModel::setShortestModel(), and usage.

ostream& operator<<	(	ostream &	ous,
		const Conseq &	cons
	)

References Conseq::cseq.

ostream& operator<<	(	ostream &	ous,
		const Conpos &	pos
	)

output frequency like the following: {A0.347526,C0.107838,G0.328182,T0.216454}

References Conpos::bases, Conpos::bc, H, and Conpos::height().

void readconf	(	const string &	file,
		map< string, string > &	dbtab,
		set< string > &	ignore,
		map< string, int > &	maxintron
	)

References dissect(), and ifstream().

Referenced by main().

void refineAllmodels	(	Connection &	conn,
		const string &	fixedfile,
		const string &	badfile,
		const string &	refinedallmod
	)

allmodels {remove bad, update stop-fixed } => allfixed

Parameters:

	refinedallmod	the result table containing allmodels: 1. without bad models. 2. stop fixed This table will not store the actually protein sequence. Only the proteinid. But the pepseq in protein table may be wrong.
	fixedfile	models whose coordinates were fixed and are good This is a subset of the allmodels. This file is loaded into a table. This schema has combined schema of model|transcript|protein, so it has a lots of columns.
	badfile	contains all bad models, they will be removed from the final modle set (refinedallmod)

Need to actually fix coordinates if they are bad, together with transcript, and protein table.

load bad models into table ///

Referenced by checkBadStopIndex().

void stoplocation	(	const string &	seq,
		vector< int > &	sloc
	)

References isStop().

Referenced by checkstop().

int stops

(

const string &

seq

)

References isStop().

void usage

(

)

given gmap summary format, this program converts it into combest archive format (*.car)

It can be used a pipe, or given specific file names.

this is a helper program to count distinct estids in the ESTId column of the combest result. So it is the actual number of ESTs mapped. This could be lower for deeply covered genome because of the coverage depth-dependent filtering.

given gmap summary format, this program converts it into combest archive format (*.car)

It can be used a pipe, or given specific file names.

this is a helper program to count distinct estids in the ESTId column of the combest result. So it is the actual number of ESTs mapped. This could be lower for deeply covered genome because of the coverage depth-dependent filtering.

template<class T>

ostream& writeBase	(	T	bf[4],
		ostream &	ous,
		const char *	sep = "\t"
	)			[inline]

References sep.

Referenced by checkstop().

ostream& writeResult	(	ostream &	ous,
		const Row &	row,
		const JGIModel &	mod
	)

References mRNAModel::printJGIModelRow(), mRNAModel::printJGITranscriptRowNoId(), and mRNAModel::proteinSequence().

Referenced by checkBadStopIndex(), and main().

ostream& writeRow	(	Row	row,
		ostream &	ous
	)