批量寻找序列中的sg结合位点

本文链接：https://blog.youkuaiyun.com/jueshengqiu/article/details/121728717

该程序用C语言实现，用于读取DNA序列文件，查找并提取序列中的sg结合位点。程序首先读取输入文件，然后通过互补配对转换，寻找'GG'和'CC'位点，并将结果写入输出文件。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

#define _CRT_SECURE_NO_WARNINGS
#pragma warning(disable: 4996)
#include <stdio.h>
#include <string.h>

void findGG(char tempGG[1001], int* Location);
void complementation(char* ptemprev, int tempsize);
void extractsg(char* temp, char* temprev, int* pGG, int* pCC);
int main()
{
   FILE *fp1 = NULL;
   FILE *fp2 = NULL;
   FILE *fp3 = NULL;
   char buff[1001] = { "" };
   char temp[1001] = { "" };
   char temprev[1001] = { "" };
   char* ptemp = NULL;
   int tempsize = 0;
   char ID[101] = { "" };
   char sg[22] = {""};
   int i = 0;
   int j = 0;
   int z = 0;
   int sgcount = 0;

int GGlocation[1001] = { 0 };
int CClocation[1001] = { 0 };

   fp1 = fopen("D:\\C\\寻找序列中的sg\\input.txt", "r");
   fp2 = fopen("D:\\C\\寻找序列中的sg\\output.txt", "w+");
   if (fp1 == NULL)
   {
       printf("fail to open the file input.txt.");
   }
   if (fp2 == NULL)
   {
       printf("fail to open the file output.txt.");
   }
   //---------------------------------------------------------------
   while (fgets(buff, 1000, fp1) != NULL)
   {
       sscanf(buff, "%s", ID);
       sscanf(buff, "%*[^\t]\t%s", temp);
       printf("%s\n", temp);
       tempsize = strlen(temp);
       strcpy(temprev, temp);
       strrev(temprev);
       complementation(temprev, tempsize);
       printf("temp\n%s\ntemprev\n%s\n", temp, temprev);
       findGG(temp, GGlocation);
       findGG(temprev, CClocation);
       extractsg(temp, temprev, GGlocation, CClocation);
       while (GGlocation[z] != 0)
       {
           GGlocation[z] = 0;
           z++;
       }
       z = 0;
       while (CClocation[z] != 0)
       {
           CClocation[z] = 0;
           z++;
       }
       z = 0;

       fp3 = fopen("D:\\C\\寻找序列中的sg\\temp.txt", "r");
       if (fp3 == NULL)
       {
           printf("fail to open the file temp.txt in main()");
       }

       i = 0;
       while (fgets(sg, 1000, fp3) != NULL)
       {
           fprintf(fp2, "%s-%d\t%s", ID, i + 1, sg);
           i++;
       }
       fclose(fp3);
   }

   fclose(fp1);
   fclose(fp2);

return 0;
}
//---------------------------------------------------------------------------------------

void findGG(char tempGG[1001], int* Location)
{
int i = 22;
int j = 0;

   while (tempGG[i] != 0)
   {
       if (tempGG[i] == 'G' || tempGG[i] == 'g')
       {
           if (tempGG[i+1] == 'G' || tempGG[i+1] == 'g')
           {
               Location[j] = i;
               printf("%d\n", i);
               j++;
           }
       }
       i++;
   }
}
//---------------------------------------------------------------------------
void complementation(char* ptemprev, int tempsize)
{
   int i = 0;
   while (i<tempsize)
   {
       if (ptemprev[i] == 'A' || ptemprev[i] == 'a')
       {
           ptemprev[i] = 'T';
       }
       else if (ptemprev[i] == 'T' || ptemprev[i] == 't')
       {
           ptemprev[i] = 'A';
       }
       else if (ptemprev[i] == 'G' || ptemprev[i] == 'g')
       {
           ptemprev[i] = 'C';
       }
       else if (ptemprev[i] == 'C' || ptemprev[i] == 'c')
       {
           ptemprev[i] = 'G';
       }
       i++;
   }
}
//---------------------------------------------------------------------------------
void extractsg(char* temp, char* temprev, int* pGG, int* pCC)
{
   int i = 0;
   int k = 0;
   FILE* fptemp = NULL;
   char sg[21] = { "" };

   fptemp = fopen("D:\\C\\寻找序列中的sg\\temp.txt", "w+");
   if (fptemp == NULL)
   {
       printf("fail to open the file temp.txt.");
   }

   while (pGG[i] != 0)
   {
       while (k<20)
       {
           sg[k] = temp[(pGG[i] - 21) + k];
           k++;
       }
       i++;
       printf("sg=%s\n", sg);
       printf("pGG[%d]=%d\n", pGG[i - 1]);
       fprintf(fptemp, "%s\n", sg);
       k = 0;
   }
   i = 0;
   k = 0;
   while (pCC[i] != 0)
   {
       while (k < 20)
       {
           sg[k] = temprev[(pCC[i] - 21) + k];
           k++;
       }
       i++;
       printf("sg=%s\n", sg);
       printf("pCC[%d]=%d\n", pCC[i - 1]);
       fprintf(fptemp, "%s\n", sg);
       k = 0;
   }
   fclose(fptemp);
}

Input.txt

Output.txt