source: cpp/frams/_demos/simil_test.cpp @ 1155

// This file is a part of Framsticks SDK.  http://www.framsticks.com/
// Copyright (C) 1999-2020  Maciej Komosinski and Szymon Ulatowski.
// See LICENSE.txt for details.


#include <vector>
#include <string>
#include "common/loggers/loggertostdout.h"
#include "frams/_demos/genotypeloader.h"
#include "frams/genetics/preconfigured.h"
#include "common/virtfile/stdiofile.h"
#include "frams/model/similarity/measure-distribution.h"
#include "frams/model/similarity/measure-greedy.h"
#include "frams/model/similarity/measure-hungarian.h"

using namespace std;

int add_double_param(std::vector<string> *args, int pos, std::vector<double> *params, std::vector<string> *params_names)
{
    for (unsigned int i = 0; i < params_names->size(); i++)
    {
        try
        {
            params->push_back(std::stod(args->at(pos)));
            pos++;
        }
        catch (const std::invalid_argument&)
        {
            printf("%s should be a number\n", params_names->at(i).c_str());
            return -1;
        }
        catch (const std::out_of_range&)
        {
            printf("%s should be inside double range\n", params_names->at(i).c_str());
            return -1;
        }
    }
    return 0;
}

/** Computes a matrix of distances between all genotypes in the specified
        .gen file, using the matching and measure weights as specified in the
        command line. */
int main(int argc, char *argv[])
{
    typedef double *pDouble;
    LoggerToStdout messages_to_stdout(LoggerBase::Enable);
    SimilMeasureBase *simil_measure = nullptr;
    if (argc < 5)
    {
        printf("Too few parameters!\n");
        printf("Command line: [-names] <genotypesFile> <measure (greedy/hungarian)> <w_dP> <w_dDEG> <w_dNEU> <w_dGEO> <fixZaxis?>\n\n");
        printf("Command line: [-names] <genotypesFile> <measure (distribution)> <desc> <simil> <dens> <bins> <samp_num>\n\n");
        printf("Parameters:\n");
        printf("  <genotypesFile> name of a file with genotypes\n");
        printf("  <measure> similarity measure name (greedy/hungarian/distribution)\n");
        printf("\n");
        printf("Parameters of greedy and hungarian measures:\n");
        printf("  <w_dP> weight of the difference in the number of parts\n");
        printf("  <w_dDEG> weight of the difference in degrees of matched parts\n");
        printf("  <w_dNEU> weight of the difference in neurons of matched parts\n");
        printf("  <w_dGEO> weight of the distance of matched parts\n");
        printf("  <fixZaxis?> should the 'z' (vertical) coordinate be fixed during the alignment? (0 or 1)\n\n");
        printf("Parameters of distribution measure:\n");
        printf("  <dens> sampling density\n");
        printf("  <bins> number of histogram bins\n");
        printf("  <samp_num> number of samples taken\n\n");

        printf("Switches:\n");
        printf("  -names specifies that the number and names of genotypes are to be printed to output\n");
        printf("  before the distance matrix; by default the number and names are not printed\n\n");

        printf("Outputs a symmetric distance matrix in the format:\n");
        printf("  <row_1> (columns in a row are separated by TABs)\n");
        printf("  ...\n");
        printf("  <row_n>\n");

        return -1;
    }

    std::vector<string> args;
    for (int i = 1; i < argc; i++)
        args.push_back(std::string(argv[i]));

    bool print_names = false;

    int pos = 1;
    if (args.at(0).compare("-names")==0)
    {
        print_names = true;
        pos = 2;
    }

    string measure_name = args.at(pos);
    pos++;
    std::vector<double> params;

    if (measure_name.compare("greedy")==0 || measure_name.compare("hungarian")==0)
    {
        std::vector<string> params_names{ "<w_dP>", "<w_dDEG>", "<w_dNEU>", "<w_dGEO>",  "<fixZaxis?>" };

        if (add_double_param(&args, pos, &params, &params_names) == -1)
            return -1;

        if (measure_name.compare("greedy")==0)
            simil_measure = new SimilMeasureGreedy();
        else
            simil_measure = new SimilMeasureHungarian();
    }


    else if (measure_name.compare("distribution")==0)
    {
        std::vector<string> params_names{ "<dens>", "<bins>", "<samp_num>" };

        if (add_double_param(&args, pos, &params, &params_names)==-1)
            return -1;

        simil_measure = new SimilMeasureDistribution();
    }

    else
    {
        printf("Measure type should be greedy (flexible criteria order and optimal matching), hungarian (vertex degree order and greedy matching) or distribution!\n");
        return -1;
    }

    simil_measure->setParams(params);

    // read the input file
    // prepare loading of genotypes from a .gen file
    // create some basic genotype converters
    PreconfiguredGenetics genetics;
    StdioFileSystem_autoselect stdiofilesys;

    // prepare output parameters from .gen file
    vector<Geno *> pvGenos;
    vector<char *> pvNames;

    long count = 0, totalsize = 0;
    GenotypeMiniLoader loader(args.at(0).c_str());
    GenotypeMini *loaded;
    while (loaded = loader.loadNextGenotype())
    {
        // while a valid genotype was loaded
        count++;
        totalsize += loaded->genotype.length();
        // create a Geno object based on the MiniGenotype
        Geno *pNextGenotype = new Geno(loaded->genotype);
        if ((pNextGenotype != NULL) && (pNextGenotype->isValid()))
        {
            pvGenos.push_back(pNextGenotype);
            char *szNewName = new char[loaded->name.length() + 1];
            strcpy(szNewName, loaded->name.c_str());
            pvNames.push_back(szNewName);
        }
        else
        {
            printf("Genotype %2li is not valid\n", count);
            if (pNextGenotype != NULL) delete pNextGenotype;
        }
    }
    if (loader.getStatus() == GenotypeMiniLoader::OnError)
    {
        printf("Error: %s", loader.getError().c_str());
    }

    double dSimilarity = 0.0;
    double **aaSimil = NULL; // array of similarities

    // create the empty array of similarities
    aaSimil = new pDouble[pvGenos.size()];
    for (unsigned int k = 0; k < pvGenos.size(); k++)
    {
        aaSimil[k] = new double[pvGenos.size()];
        for (unsigned int l = 0; l < pvGenos.size(); l++)
            aaSimil[k][l] = 0.0;
    }



    // compute and remember similarities
    for (unsigned int i = 0; i < pvGenos.size(); i++)
    {
        for (unsigned int j = 0; j < pvGenos.size(); j++)
        {
            dSimilarity = simil_measure->evaluateDistance(pvGenos.operator[](i), pvGenos.operator[](j));
            aaSimil[i][j] = dSimilarity;
        }
    }

    if (print_names)
    {
        // if "-names" switch was given, print the number of genotypes and their names
        printf("%li\n", pvGenos.size());
        for (unsigned int iGen = 0; iGen < pvNames.size(); iGen++)
        {
            printf("%s\n", pvNames.at(iGen));
        }
    }

    // print out the matrix of similarities
    for (unsigned int i = 0; i < pvGenos.size(); i++)
    {
        for (unsigned int j = 0; j < pvGenos.size(); j++)
        {
            printf("%.2lf\t", aaSimil[i][j]);
        }
        printf("\n");
    }

    // delete vectors and arrays
    for (unsigned int i = 0; i < pvGenos.size(); i++)
    {
        delete pvGenos.operator[](i);
        delete[] pvNames.operator[](i);
        delete[] aaSimil[i];
    }

    delete[] aaSimil;
    delete simil_measure;

    return 0;
}