DeleteEndsInGPD.cpp

#include <iostream>
#include <string>
#include <sstream>
#include <fstream>
#include <vector>
#include <cstdlib>
#include <algorithm>
#include <boost/algorithm/string/split.hpp>
#include <boost/algorithm/string.hpp>
#include <boost/lexical_cast.hpp>


using namespace std;
using namespace boost;


struct GPD {

    string gname;
    string tname;
    string chro;
    string strand;
    int start;
    int end;
    int cstart;
    int cend;
    int nblock;
    vector<int> vec_exon_start;
    vector<int> vec_exon_end;
};


void BuildGPDFromString (string &str, GPD &gpd) {

    vector<string> vec_temp;
    boost::split(vec_temp, str, boost::is_any_of("\t"));
    gpd.gname = vec_temp[0];
    gpd.tname = vec_temp[1];
    int L = vec_temp[2].length();
    gpd.chro = vec_temp[2];
    gpd.strand = vec_temp[3];
    gpd.start = lexical_cast<int>(vec_temp[4]);
    gpd.end = lexical_cast<int>(vec_temp[5]);
    gpd.cstart = lexical_cast<int>(vec_temp[6]);
    gpd.cend = lexical_cast<int>(vec_temp[7]);
    gpd.nblock = lexical_cast<int>(vec_temp[8]);
    vector<string> vec1;
    boost::split(vec1, vec_temp[9], boost::is_any_of(","));
    for(int i = 0; i < vec1.size() - 1; ++i)  (gpd.vec_exon_start).push_back(lexical_cast<int>(vec1[i]));
    vector<string> vec2;
    boost::split(vec2, vec_temp[10], boost::is_any_of(","));
    for(int i = 0; i < vec2.size() - 1; ++i)  (gpd.vec_exon_end).push_back(lexical_cast<int>(vec2[i]));
}




int main (int argc, char **argv) {

    ifstream inf(argv[1]);

    while(inf){

        string strInput;
        getline(inf, strInput);

        if(strInput.length() > 0){

            GPD gpd;
            BuildGPDFromString(strInput, gpd);

            if(gpd.nblock > 1){

                cout << gpd.gname << '\t' << gpd.tname << '\t' << gpd.chro << '\t' << gpd.strand << '\t' << gpd.nblock << '\t';
                for(int i = 1; i < gpd.nblock; ++i)  cout << gpd.vec_exon_start[i] << ',';
                cout << '\t';
                for(int i = 0; i < gpd.nblock - 1; ++i)  cout << gpd.vec_exon_end[i] << ',';
                cout << endl;
            }
        }
    }

    inf.close();

    return 0;
}