api_gene_name.cpp

/*Take a file.txt (list of specific gene he wants to study from a researcher) and a file.tsv (gene present in the matrix table)
 * and returns a set with all the different possible names of these genes that are also present in the matrix table
 *
 * The API will work like that: read gene.tsv file, stores all gene names in a set, 
 * then read gene.txt file, store in a vector, 
 * then call the API on this vector to create a set with all the possible names of the genes, 
 * and then take the intersection of the 2 sets (so that we only keep the one we are interested in), 
 * transform the set into a vector and return this vector. 
 */

#include "api_gene_name.h"
#include "readgenetxt.h"
#include "only_gene_name.h"
#include "read_tsv_set.h"
#include "intersection_sets.h"
#include "api.h"

#include <iostream>
#include <string>
#include <sstream>
#include <iterator>
#include <vector>
#include <algorithm>
#include <fstream>
#include <QJsonDocument>
#include <stdio.h>
api_gene_name::api_gene_name()
{

}

std::vector<std::string> api_gene_name::api_gene_name_funtion(std::string geneNameFile, std::string geneSubsetFile){

    //first we open the set of the genes the researcher is interested and call the
    //API on each gene to find all their different names
    std::cout<<"entered gene_name_function"<<std::endl;
    readgenetxt vec;
    std::vector<std::string> res;
    res=vec.listgene(geneSubsetFile); //list of all the genes in the txt file
    
    CURL *curl = curl_easy_init();

    std::set<std::string> final_set;
    for (unsigned int i=0; i<res.size();i++){
        std::cout<<"loop number: "<<i<<std::endl;
        std::string search=res[i];

        //API call

        std::string l = searchHGNC(search, curl);

        only_gene_name test; //search other name of this specific gene in the string l
        std::set<std::string> small_set;
        small_set=test.listgene(l, search);

        final_set.insert(small_set.begin(), small_set.end()); //add the set of the names of this gene in the set of all the genes

    }
    curl_easy_cleanup(curl);
    //Now let us create the set of all the gene of the matrix file
    read_tsv_set test2;
    std::set<std::string> string_set_gene_matrix;
    string_set_gene_matrix=test2.read_tsv(geneNameFile);

    //Now we take the intersection of the 2 sets, to keep only the genes we are interrested in
    intersection_sets test3;
    std::set<std::string> intersection_set;
    intersection_set = test3.set_intersection(final_set, string_set_gene_matrix);
    
    std::cout << std::endl;
    std::cout<<"[Progress API gene name finished], Intersection set is: "<<std::endl;
    for (std::string x : intersection_set) {
        std::cout << x << " ";
    }    

    std::vector<std::string> v = std::vector<std::string>();

    for (std::string x : intersection_set) {
        std::cout << x << " ";
        v.push_back(x);
    }
    std::cout << std::endl;

    std::cout << "Set operations finished" << std::endl;


    return v;
    }