//SequenceFinder.cpp
//Coded by Vishakh (vishakh@uci.edu)
//ICS 175A, Winter 2004
//Project: Motif detection in yeast
//Purpose: Calculates scores for substrings of gene families, finds sequences and constructs
//probability matrices.

#include<iostream>
#include<cmath>
#include<algorithm>
#include "dtable.h"
#include "dstring.h"
#include "ProbMatrix.h"
#include "Permutation.h"

using namespace std;

struct MatrixPair
{
	DString name;
	double score;
};

class SequenceFinder
{
public:
	SequenceFinder();
	void setKmerSize(int n);
	void setOccurenceLimit(int n);
	void assignScores(DTable& dt_big, DTable& dt_sml);
	void HTMLOutput(int numOfKmers, char *mainPageName);
	void sortTable();
	void printTopScores(int n);
	void createMatrices(vector<vector<GenomicPair> > genes);
	vector<GenomicPair> getTable();
	vector<ProbMatrix> getMatrices();

	// math functions defined
	double poissonDistribution(double lambda, double poisson_k);
	double binomialDistribution(double bin_k, double bin_n, double bin_p);
	double inverseLog10(double log);
	double logOfFactorial(double n);
	double factorial(double n);

private:
	vector<GenomicPair> table_sml;
	vector<GenomicPair> table_big;
	vector<ProbMatrix> matrices;

	int KMER_SIZE;
	int OCCURENCE_LIMIT;
};