dts_pss_crf.cpp

用于蛋白质数据生成分类数据的转换。做PSS（Protein Secondary Structure prediction)的可以参考一下。

// dts_pss_crf.cpp : Defines the entry point for the console application.
//
#include <string>
#include <iostream>
#include <fstream>
#include <sstream>
#include <vector>
using namespace std;

#define WIDTH 7

int main(int argc, char* argv[])
{
        ifstream in;
        if(argc!=3){
                cout<<"usage .exe infile outfile"<<endl;
                return -1;
        }
        in.open(argv[1]);
        string acid;
        getline(in,acid);
        string structure;
        getline(in,structure);
        getline(in,structure);
        getline(in,structure);
        getline(in,structure);
        getline(in,structure);


        acid = acid.substr(4,acid.size());
        structure = structure.substr(7,structure.size());

        vector<char> acid_string;
        vector<char> structure_string;
        size_t i;
        for(i=0;i<acid.size();i+=2){
                acid_string.push_back(acid[i]);
        }
        for(i=0;i<structure.size();i+=2){
                structure_string.push_back(structure[i]);
        }

        for(i=0;i<acid_string.size();i++)
                cout<<acid_string[i]<<" ";
        cout<<endl;
        for(i=0;i<structure_string.size();i++)
                cout<<structure_string[i]<<" ";
        cout<<endl;
		ofstream out;
		out.open(argv[2]);
		for(i=0;i<structure_string.size();i++){
			out<<acid_string[i]<<" "<<structure_string[i]<<endl;
		}
		out<<endl;
		
}