rmfalselink.cpp File Reference

#include <iostream>
#include <iomanip>
#include <fstream>
#include "mysql++.h"
#include <map>
#include "DBStat.h"
#include <gsl/gsl_cdf.h>
#include <cmath>

Classes
class	FitNgidentity
struct	Resrow
struct	Intabcol
Functions
void	readStat (Connection &conn, const string &table, map< string, map< string, SNavgstd > > &dbindex)
Dbstat *	readStatInt (Connection &conn, const string &table)
int	removeFalsePositiveHits (Connection &conn, const string &stattab, Query &query, const string &outfile, vector< Resrow > &blk)
void	storeBadInTable (Connection &conn, const vector< Resrow > &lk, const string &tab, const float lowestng)
int	main (int argc, char *argv[])
double	chisquare (const double *obsv, const double *expt, const int n, double *&zv)

---

Function Documentation

double chisquare	(	const double *	obsv,
		const double *	expt,
		const int	n,
		double *&	zv
	)

check the two arrays, ignore zero values in obsv obsv has n+1 elements, element[0] was not nused. the index number is the dbid that start from 1.

Parameters:

expt

is an array containing two elements mean,std encoded in (2i-1, 2i), with i=1, 2, ..., 2n the array has 2n+1 elements, element at zero index was not used so that the index can start from 1.

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

References Intabcol::db1id, Intabcol::db1seqid, Intabcol::db2id, Intabcol::db2seqid, removeFalsePositiveHits(), storeBadInTable(), and user.

void readStat	(	Connection &	conn,
		const string &	table,
		map< string, map< string, SNavgstd > > &	dbindex
	)

the table structure for stat +-----------+---------------+------+-----+---------+-------+ | Field | Type | Null | Key | Default | Extra | +-----------+---------------+------+-----+---------+-------+ | db1 | varchar(30) | NO | | NULL | | | db2 | varchar(30) | NO | | NULL | | | avgscore | decimal(14,4) | YES | | NULL | | | stdscore | double(26,4) | YES | | NULL | | | avgiden | decimal(18,8) | YES | | NULL | | | stdiden | double(34,8) | YES | | NULL | | | avgngiden | decimal(18,8) | YES | | NULL | | | stdngiden | double(34,8) | YES | | NULL | | +-----------+---------------+------+-----+---------+-------+

This is the string version. For better performance I am implementing the array version.

References string().

Dbstat * readStatInt	(	Connection &	conn,
		const string &	table
	)

use integer index instead of string map

References Dbstat::add(), and FitNgidentity::numdb.

Referenced by removeFalsePositiveHits().

int removeFalsePositiveHits	(	Connection &	conn,
		const string &	stattab,
		Query &	query,
		const string &	outfile,
		vector< Resrow > &	blk
	)

stat table should have the following definition: mysql:kemin:shake>desc db2dbngstat95; +-----------+------------------+------+-----+---------+-------+ | Field | Type | Null | Key | Default | Extra | +-----------+------------------+------+-----+---------+-------+ | db1 | int(10) unsigned | NO | PRI | NULL | | | db2 | int(10) unsigned | NO | PRI | NULL | | | avgscore | decimal(14,4) | YES | | NULL | | | stdscore | double(26,4) | YES | | NULL | | | avgiden | decimal(18,8) | YES | | NULL | | | stdiden | double(34,8) | YES | | NULL | | | avgngiden | decimal(18,8) | YES | | NULL | | | stdngiden | double(34,8) | YES | | NULL | | | count | bigint(21) | NO | | 0 | | +-----------+------------------+------+-----+---------+-------+

The sqlstr should pull the following columns query << "select db1, seq1name, db2, seq2name, score, ngidentity from " << inputtable << " order by db1, seq1name, db2 ";

This function will be able to work on both the forward and backward direction. The backward has multiple hits from the forward which we picked only one hit. So the backward direction should use the max ngidentity. We are elminating the hits if found lower that other databases. Since we have done a 95% cutoff, this should not be a problem.

taken information from the raw data and do some test

References Dbstat::getNumberOfDb(), Dbstat::getStatRow(), and readStatInt().

Referenced by main().

void storeBadInTable	(	Connection &	conn,
		const vector< Resrow > &	lk,
		const string &	tab,
		const float	lowestng
	)

The output table has the fixed schema: db1 integer, seq1id integer unsigned, db2 integer, ngidentity float, primary key(db1,seq1id,db2))

References seq1.

Referenced by main().