roc.c

C-package of "Long Short-Term Memory" for Protein classification

/*
 * $Id: roc.c 1269 2007-05-09 07:49:08Z mhe $
 */

#include <stdio.h>
#include <stdlib.h>
#include "lstm.h"
#include "roc.h"

/* compares two seqperfs */
int seqperfcmp(const void *sp0, const void *sp1) {
    
    if ((*(struct seqperf*)sp0).out == ((*(struct seqperf*)sp1)).out) {
        return 0;
    }
    else if ((*(struct seqperf*)sp0).out > (*(struct seqperf*)sp1).out) {
        return -1;
    }
    else return 1;
    
}

static double *xarrtest, *yarrtest, *xarrtrain, *yarrtrain;

void allocarrs(int train_size, int test_size) {
    
    xarrtest = (double*)malloc(sizeof(double) * test_size);
    yarrtest = (double*)malloc(sizeof(double) * test_size);
    xarrtrain = (double*)malloc(sizeof(double) * train_size);
    yarrtrain = (double*)malloc(sizeof(double) * train_size);
    
}

double roctest(int num_prot) {
    
    int i, fp, tp;
    double auc, prevx;
    
    qsort(sptest, num_prot, sizeof(struct seqperf), *seqperfcmp);
    
    tp = 0;
    fp = 0;
    
    for (i = 0; i < num_prot; i++) {
        
        if (sptest[i].target == 1.0) {
            tp++;
        }
        else {
            fp++;
        }
        xarrtest[i] = (double) fp / (negativetest * 1.0);
        yarrtest[i] = (double) tp / (positivetest * 1.0);
    
    }
    
    auc   = 0;
    prevx = 0;
    for (i = 0; i < num_prot; i++) {

        if (xarrtest[i] != prevx) {
            auc += (xarrtest[i] - prevx) * yarrtest[i];
            prevx = xarrtest[i];
        }

    }
  
    return auc;
}

double roctrain(int num_prot) {
    
    int i, fp, tp;
    double auc, prevx;
    
    qsort(sptrain, num_prot, sizeof(struct seqperf), *seqperfcmp);
    
    tp = 0;
    fp = 0;
    
    for (i = 0; i < num_prot; i++) {
        
        if (sptrain[i].target == 1.0) {
            tp++;
        }
        else {
            fp++;
        }
        xarrtrain[i] = (double) fp / (negativetrain * 1.0);
        yarrtrain[i] = (double) tp / (positivetrain * 1.0);
    
    }
    
    auc   = 0;
    prevx = 0;
    for (i = 0; i < num_prot; i++) {

        if (xarrtrain[i] != prevx) {
            auc += (xarrtrain[i] - prevx) * yarrtrain[i];
            prevx = xarrtrain[i];
        }

    }
  
    return auc;
}

double rocROCNtest(int num_prot) {
    
    int i, j, fp, tp;
    double auc, prevx;
    
    qsort(sptest, num_prot, sizeof(struct seqperf), *seqperfcmp);
    
    tp = 0;
    fp = 0;
    
    for (i = 0; i < num_prot; i++) {
        
        if (sptest[i].target == 1.0) {
            tp++;
        }
        else {
            fp++;
        }
        xarrtest[i] = (double) fp / (double)ROCN;
        yarrtest[i] = (double) tp / (positivetest * 1.0);
        
        if (fp == ROCN) {
            break;
        }
    
    }
    
    auc   = 0;
    prevx = 0;
    for (j = 0; j <= i; j++) {

        if (xarrtest[j] != prevx) {
            auc += (xarrtest[j] - prevx) * yarrtest[j];
            prevx = xarrtest[j];
        }

    }
  
    return auc;
}

double rocROCNtrain(int num_prot) {
    
    int i, j, fp, tp;
    double auc, prevx;
    
    qsort(sptrain, num_prot, sizeof(struct seqperf), *seqperfcmp);
    
    tp = 0;
    fp = 0;
    
    for (i = 0; i < num_prot; i++) {
        
        if (sptrain[i].target == 1.0) {
            tp++;
        }
        else {
            fp++;
        }
        xarrtrain[i] = (double) fp / (double)ROCN;
        yarrtrain[i] = (double) tp / (positivetrain * 1.0);
        
        if (fp == ROCN) {
            break;
        }
    
    }
    
    auc   = 0;
    prevx = 0;
    for (j = 0; j <= i; j++) {

        if (xarrtrain[j] != prevx) {
            auc += (xarrtrain[j] - prevx) * yarrtrain[j];
            prevx = xarrtrain[j];
        }

    }
  
    return auc;
}