Notes

#include <stdio.h>
#include <string.h>


int SearchStringInArray( char* array[], char* key, int size )
{
int n;
for ( n = 0; n < size; ++n ) {
if ( !strcmp(array[n], key) ) {

return n;
}
}
return -1;

}

int main(int argc,char *argv[]) {
char *fenilalanin[2] = {"UUU", "UUC"};
char *losin[6] = {"UUA", "UUG", "CUU", "CUC", "CUA", "CUG"};
char *izolosin[3] = {"AUU", "AUC", "AUA"};
char *metionin[1] = {"AUG"};
char *valin[4] = {"GUU", "GUC", "GUA", "GUG"};
char *serin[6] = {"UCU", "UCC", "UCA", "UCG", "AGU", "AGC"};
char *prolin[4] = {"CCU", "CCC", "CCA", "CCG"};
char *treonin[4] = {"ACU", "ACC", "ACA", "ACG"};
char *alanin[4] = {"GCU", "GCC", "GCA", "GCG"};
char *triozin[2] = {"UAU", "UAC"};
char *stop_codon[3] = {"UAA", "UAG", "UGA"};
char *histidin[2] = {"CAU", "CAC"};
char *glutamin[2] = {"CAA", "CAG"};
char *asparajin[2] = {"AAU", "AAC"};
char *lizin[2] = {"AAA", "AAG"};
char *aspartik_acid[2] = {"GAU", "GAC"};
char *glutamic_acid[2] = {"GAA", "GAG"};
char *sistein[2] = {"UGU", "UGC"};
char *triptofan[1] = {"UGG"};
char *arjinin[6] = {"CGU", "CGC", "CGA", "CGG", "AGA", "AGG"};
char *glisin[4] = {"GGU", "GGC", "GGA", "GGG"};
FILE *fp;
char protein[255];
fp = fopen(argv[1], "r");
fscanf(fp, "%[^n]", protein);
printf("fname: %sFile content: %sn",argv[1],protein );
printf("n") ;
char str1[4] ;
char str2[4] ;
char str3[4] ;
char str4[4] ;
char str5[4] ;
char str6[4] ;
char str7[4] ;
char str8[4] ;
char str9[4] ;
char str10[4] ;
memcpy(str1,protein, 3);
memcpy(str2, protein+3, 3);
memcpy(str3, protein+6, 3);
memcpy(str4, protein+9, 3);
memcpy(str5, protein+12, 3);
memcpy(str6, protein+15, 3);
memcpy(str7, protein+18, 3);
memcpy(str8, protein+21, 3);
memcpy(str9, protein+24, 3);
memcpy(str10, protein+27, 3);
char start[] = "AUG";
if (strcmp(str1,start) == 0 && ( strcmp(str10,stop_codon[0])== 0 || strcmp(str10,stop_codon[1])== 0 || strcmp(str10,stop_codon[2])== 0 ) ) {
if (SearchStringInArray(valin, str2, 4) != -1 && SearchStringInArray(alanin, str3, 4) != -1 &&
SearchStringInArray(glutamic_acid, str4, 2) != -1 &&
SearchStringInArray(glisin, str5, 4) != -1 &&
SearchStringInArray(treonin, str6, 4) != -1 &&
SearchStringInArray(lizin, str7, 2) != -1 &&
SearchStringInArray(arjinin, str8, 6) != -1 &&
SearchStringInArray(izolosin, str9, 3) != -1 &&
SearchStringInArray(stop_codon, str10, 3) != -1 )
{
printf("MyProteinA is identified in sequence.n");
printf("The amino acids of MyProteinA: M-V-A-E-G-T-K-R-In") ;
printf("proteinA: M-V-A-E-G-T-K-R-In") ;
}
else if ((SearchStringInArray(glisin, str2, 4) != -1) && (SearchStringInArray(glutamic_acid, str3, 2) != -1) &&
(SearchStringInArray(alanin, str4, 4) != -1) &&
(SearchStringInArray(valin, str5, 4) != -1) &&
(SearchStringInArray(arjinin, str6, 6) != -1) &&
(SearchStringInArray(lizin, str7, 2) != -1) &&
(SearchStringInArray(treonin, str8, 4) != -1) &&
(SearchStringInArray(izolosin, str9, 3) != -1) &&
(SearchStringInArray(stop_codon, str10, 3) != -1) )
{
printf("MyProteinB is identified in sequence.n") ;
printf("The amino acids of MyProteinB: M-G-E-A-V-R-K-T-In") ;
printf("proteinB: M-G-E-A-V-R-K-T-In") ;
}
else if ((SearchStringInArray(fenilalanin, str2, 2) != -1) && (SearchStringInArray(serin, str3, 6) != -1) &&
(SearchStringInArray(triozin, str4, 2) != -1) &&
(SearchStringInArray(sistein, str5, 2) != -1) &&
(SearchStringInArray(losin, str6, 6) != -1) &&
(SearchStringInArray(prolin, str7, 4) != -1) &&
(SearchStringInArray(glutamin, str8, 2) != -1) &&
(SearchStringInArray(arjinin, str9, 6) != -1) &&
(SearchStringInArray(stop_codon, str10, 3) != -1) )
{
printf("MyProteinC is identified in sequence.n") ;
printf("The aminoacids of MyProteinC: M-F-S-Y-C-L-P-Q-Rn");
printf("proteinC: M-F-S-Y-C-L-P-Q-Rn") ;
}
else if ((SearchStringInArray(fenilalanin, str2, 2) != -1) && (SearchStringInArray(losin, str3, 6) != -1) &&
(SearchStringInArray(valin, str4, 4) != -1) &&
(SearchStringInArray(prolin, str5, 4) != -1) &&
(SearchStringInArray(treonin, str6, 4) != -1) &&
(SearchStringInArray(triozin, str7, 2) != -1) &&
(SearchStringInArray(aspartik_acid, str8, 2) != -1) &&
(SearchStringInArray(histidin, str9, 2) != -1) &&
(SearchStringInArray(stop_codon, str10, 3) != -1) )
{
printf("MyProteinD is identified in sequence.n") ;
printf("The aminoacids of MyProteinD: M-F-L-V-P-T-Y-D-Hn") ;
printf("proteinD: M-F-L-V-P-T-Y-D-Hn");
}
else if ((SearchStringInArray(fenilalanin, str2, 2) != -1) && (SearchStringInArray(serin, str3, 6) != -1) &&
(SearchStringInArray(triozin, str4, 2) != -1) &&
(SearchStringInArray(sistein, str5, 2) != -1) &&
(SearchStringInArray(losin, str6, 6) != -1) &&
(SearchStringInArray(prolin, str7, 4) != -1) &&
(SearchStringInArray(lizin, str8, 2) != -1) &&
(SearchStringInArray(arjinin, str9, 6) != -1) &&
(SearchStringInArray(stop_codon, str10, 3) != -1) )
{
printf("MyProteinE is identified in sequence.n") ;
printf("The aminoacids of MyProteinE: M-F-S-Y-C-L-P-K-Rn") ;
printf("proteinE: M-F-S-Y-C-L-P-K-Rn") ;
}
else {
printf("It is not a known proteinn") ;
printf("It is probably a new protein. Mn") ;
}


}

else {
if (strcmp(str1,start)!= 0)
printf("Not start with AUG it is not a proteinn");
else
printf("It is not a protein! No Stop coden") ;
}
return 0;
}