dnaSequence.c

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define NO_FILE_ERR NULL
#define NO_DATA_ERR NULL

#ifndef DEBUG
    #undef DEBUG
    #define DEBUG
    #define SEVERITY 2
#endif
char *loadSequenceN(const char *alphabet);
int printSequenceV(const char *sequence, int from, const int to);
int sequenceHistogramH(const char *sequence);
int subSequenceFinderP();
char *insertSequenceL(const int pos, const int length, const char *newSequence, const char *sequence);
int longestSubstrInSequenceO();
char *substring(char *string, int position, int length);
int contains(const char needle, const char *haystack);
int isLegit(const char *sequence, const char *alphabet, const int size);

int main()
{
    //FILE *f = fopen("DNAsekvencia.txt", "r");
    char *sequence = NULL;
    char newSequence[1000];
    char *alphabet = "ACGTacgt";
    char selection = NULL;
    int from = 0, to = 0;
    /*if (f == NULL)
    {
        printf("Neotvoreny subor\n");
        return NO_FILE_ERR;
    }*/
    while (selection != 'k')
    {
        selection = getchar();
        if (selection == 'n')
        {
            sequence = loadSequenceN(alphabet);
    #ifdef DEBUG
        #if SEVERITY >= 1
            printf("loadSequenceN() return value on line %d: %s\n", __LINE__, sequence);
        #endif
    #endif
        }
        else if (selection == 'v')
        {
            scanf("%d %d", &from, &to);
            printSequenceV(sequence, from, to);
            getchar();
        }
        else if (selection == 'h')
        {
            sequenceHistogramH(sequence);
        }
        else if (selection == 'p')
        {
            subSequenceFinderP();
        }
        else if (selection == 'l')
        {
            scanf("%d %d %s", &from, &to, newSequence);
            sequence = insertSequenceL(from, to, newSequence, sequence);
        }
        else if (selection == 'o')
        {
            longestSubstrInSequenceO(sequence);
        }
    }

    return 0;
}

char *loadSequenceN(const char *alphabet)
{
    FILE *f = fopen("DNAsekvencia.txt", "r");
    //TODO finish if there is stuff changed...or maybe not, 'cause it works like charm :D
    int index = 0;
    int size = 0;
    char c;
    char *sequence = NULL;

    rewind(f);

    if (f == NO_FILE_ERR)
    {
        printf("Neotvoreny subor\n");
        return NO_FILE_ERR;
    }

    //file length checking
    fseek( f , 0L , SEEK_END);
    size = ftell( f );
    rewind( f );

    //allocating the string
    sequence = (char *)malloc((size + 1) * sizeof(char));
    fgets (sequence, size+1, f);

    fclose(f);
    #ifdef DEBUG
        #if SEVERITY >= 1
            printf("loadSequenceN() on line %d: %s\n", __LINE__, sequence);
        #endif
    #endif
    if(isLegit(sequence, alphabet, size))
    {
        printf("Sekvenciu sa podarilo nacitat\n");
        return sequence;
    }
    return NULL;
}

int printSequenceV(const char *sequence, int from, const int to)
{
    if(sequence == NO_DATA_ERR)
    {
        printf("Sekvencia nie je nacitana\n");
        return NO_DATA_ERR;
    }
    if(from < 1 || to < 1 || from > to)
    {
        printf("Zadanu podsekvenciu nie je mozne vypisat\n");
        return 1;
    }
    for(from; from <= to; from++)
    {
        printf("%c", sequence[from - 1]);
    }
    printf("\n");
    return 0;
}

int sequenceHistogramH(const char *sequence)
{
    int aCount = 0;
    int cCount = 0;
    int gCount = 0;
    int tCount = 0;
    int i = 0;
    for(i; i < strlen(sequence); i++)
    {
        if(sequence[i] == 'A' || sequence[i] == 'a')
        {
            aCount++;
    #ifdef DEBUG
        #if SEVERITY >= 2
            printf("From sequenceHistogramH(), var aCount on line %d: %d\n", __LINE__, aCount);
        #endif
    #endif
            continue;
        }
        if(sequence[i] == 'C' || sequence[i] == 'c')
        {
            cCount++;
    #ifdef DEBUG
        #if SEVERITY >= 2
            printf("From sequenceHistogramH(), var cCount on line %d: %d\n", __LINE__, cCount);
        #endif
    #endif
            continue;
        }
        if(sequence[i] == 'G' || sequence[i] == 'g')
        {
            gCount++;
    #ifdef DEBUG
        #if SEVERITY >= 2
            printf("From sequenceHistogramH(), var gCount on line %d: %d\n", __LINE__, gCount);
        #endif
    #endif
            continue;
        }
        if(sequence[i] == 'T' || sequence[i] == 't')
        {
            tCount++;
    #ifdef DEBUG
        #if SEVERITY >= 2
            printf("From sequenceHistogramH(), var tCount on line %d: %d\n", __LINE__, tCount);
        #endif
    #endif
            continue;
        }
    }

    printf("%c: %d\n%c: %d\n%c: %d\n%c: %d\n", 'A', aCount, 'C', cCount, 'G', gCount, 'T', tCount);
    return 0;
}

int subSequenceFinderP()
{
    printf("subSequence\n");
    return 0;
}

char *insertSequenceL(int pos, const int length, const char *newSequence, const char *sequence)
{
    char *placeholder = NULL;
    int size = (strlen(newSequence) + strlen(sequence) + 2);

    if(sequence == NO_DATA_ERR)
    {
        printf("Sekvencia nie je nacitana\n");
        return sequence;
    }

    if(pos > strlen(sequence) + 1 || pos < 1)
    {
        printf("Do sekvencie nie je mozne podsekvenciu vlozit\n");
        return sequence;
    }

    placeholder = (char *)malloc(size * sizeof(char));
    memset(placeholder, 0, size);

    pos--;
    // TODO complete this
    strncpy(placeholder, sequence, pos);
    int len = strlen(placeholder);
    strcpy(placeholder+len, newSequence);
    len += strlen(newSequence);
    strcpy(placeholder+len, sequence+pos);
    printf("Podsekvencia vlozena\n");
    return placeholder;

}

int longestSubstrInSequenceO(const char *str)
{
    int i = 0;
    char *cntrlStr = (char *)malloc((strlen(str)) * sizeof(char));
    char *copied  = (char *)malloc((strlen(str)) * sizeof(char));
    int size;
    strcpy(copied, str);

    for ( ; *copied; ++copied)
    {
        *copied = tolower(*copied);
    }

    for(i; i < strlen(str); i++)
    {
        printf("%d\n", (strlen(str) - i + 1));

        memset(cntrlStr, 'a', (strlen(str) - i ) * sizeof(char));
        printf("%s\n", cntrlStr);
        if(strstr(copied, cntrlStr) != NULL)
        {
            printf("Longest sequence: %d\n", i);
            break;
        }
        memset(cntrlStr, 'c', (strlen(str) - i) * sizeof(char));
        printf("%s\n", cntrlStr);
        if(strstr(copied, cntrlStr) != NULL)
        {
            printf("Longest sequence: %d\n", i);
            break;
        }
        memset(cntrlStr, 'g', (strlen(str) - i) * sizeof(char));
        printf("%s\n", cntrlStr);
        if(strstr(copied, cntrlStr) != NULL)
        {
            printf("Longest sequence: %d\n", i);
            break;
        }
        memset(cntrlStr, 't', (strlen(str) - i) * sizeof(char));
        printf("%s\n", cntrlStr);
        if(strstr(copied, cntrlStr) != NULL)
        {
            printf("Longest sequence: %d\n", i);
            break;
        }
        memset(cntrlStr, 0, (strlen(str) - i) * sizeof(char));
        printf("%s\n", cntrlStr);

        free(cntrlStr);
        cntrlStr = NULL;
        cntrlStr = (char *)malloc((strlen(str) - i ) * sizeof(char));
        //memset(cntrlStr, 0, (strlen(str) - i ) * sizeof(char));

    }
    printf("search end\n");
    //strstr();
    return 0;
}

char *substring(char *string, int position, int length)
{
   char *pointer;
   int c;

   pointer = (char *)malloc((length+1) * sizeof(char));
   if( pointer == NULL )
       exit(EXIT_FAILURE);

   for( c = 0 ; c < length ; c++ )
      *(pointer+c) = *((string+position-1)+c);

   *(pointer+c) = '\0';

   return pointer;
}

int contains(const char needle, const char *haystack)
{
    char check = NULL;
    check = strchr(haystack, (int)needle);
    #ifdef DEBUG
        #if SEVERITY >= 3
            printf("From contains() on line %d: %c\n", __LINE__, check);
        #endif
    #endif
    if(check == NULL)
    {
        return 0;
    }
    else
    {
        return 1;
    }
}

int isLegit(const char *sequence, const char *alphabet, const int size)
{
    int index = 0;
    for(index; index < size; index++)
    {
        if(contains(sequence[index], alphabet))
        {
    #ifdef DEBUG
        #if SEVERITY >= 2
            printf("From isLegit() on line %d: %c\n", __LINE__, sequence[index]);
        #endif
    #endif
            continue;
        }
        else
        {
            printf("Sekvencia nesplna podmienky\n");
            return 0;
        }
    }
    return 1;
}