dbaln_nr.cpp File Reference

#include "mysql++.h"
#include <string>
#include "bioseq.h"
#include "dynaln.h"
#include <math.h>
#include <fstream>
#include "strformat.h"
#include <map>
#include "dbinfo.h"
#include "dbalnhelper.h"

Defines
#define	MYSQLPLUS2
Functions
void	getPrt (mysqlpp::Query &q, const string &id, string &prt)
void	alignProtein (const string &alnfile, const string &alntabfile, const string &pairtab, const string &prtsrctab, Connection &conn, const int bidx, const int rowcnt)
void	alignDNA (const string &alnfile, const string &alntabfile, const string &pairtab, mysqlpp::Connection &conn, mysqlpp::Connection &conn1, mysqlpp::Connection &conn2)
void	bufferGenomic (mysqlpp::Connection &conn, const string &sqlstr, map< string, DNA > &id2seq)
void	storeProteinIntoMap (Connection &conn, const string &srctab, map< unsigned int, string > &id2seq, const string &pairtab, const int bidx, const int rowcnt)
void	storeInMap (Connection &conn, const string &srctab, map< unsigned int, string > &id2seq, const string &pairtab, const int bidx, const int rowcnt)
int	main (int argc, char *argv[])

---

Define Documentation

#define MYSQLPLUS2

---

Function Documentation

void alignDNA	(	const string &	alnfile,
		const string &	alntabfile,
		const string &	pairtab,
		mysqlpp::Connection &	conn,
		mysqlpp::Connection &	conn1,
		mysqlpp::Connection &	conn2
	)

void alignProtein	(	const string &	alnfile,
		const string &	alntabfile,
		const string &	pairtab,
		const string &	prtsrctab,
		Connection &	conn,
		const int	bidx,
		const int	rowcnt
	)

void bufferGenomic	(	mysqlpp::Connection &	conn,
		const string &	sqlstr,
		map< string, DNA > &	id2seq
	)

void getPrt	(	mysqlpp::Query &	q,
		const string &	id,
		string &	prt
	)

DESCRIPTION use three connections, the first connection provide the alignment pairs. the second and the third provided the raw sequences All three connections can be the same. each connection provide a table.

Input: all from one database the tab must have id and seq column

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

References alignProtein(), MysqlDBInfo::getAuthenInfo(), MysqlDBInfo::getPassword(), and MysqlDBInfo::getUser().

void storeInMap	(	Connection &	conn,
		const string &	srctab,
		map< unsigned int, string > &	id2seq,
		const string &	pairtab,
		const int	bidx,
		const int	rowcnt
	)

this take advantage the fact that all primary key in the nr should be the gi number which is a value < 2 billion so far

void storeProteinIntoMap	(	Connection &	conn,
		const string &	srctab,
		map< unsigned int, string > &	id2seq,
		const string &	pairtab,
		const int	bidx,
		const int	rowcnt
	)

this function fetches protein sequences and stored them locally as a map data structure. Using a tree map is very slow. needs to use a hash_map