c_cpp/libgFileIO/chrseq.h

   1 #ifndef _G_CHRSEQ_H
   2 #define _G_CHRSEQ_H
   3
   4 #include <stdint.h>     // int_fast8_t
   5 #include <stdlib.h> // malloc
   6
   7 FORCE_INLINE void NormalizeChrSeq(char *seq){
   8     while(*seq){
   9         switch (*seq) {
  10             case 'a': case 'A':
  11                     *seq = 'A';
  12                     break;
  13             case 't': case 'T':
  14                     *seq = 'T';
  15                     break;
  16             case 'c': case 'C':
  17                     *seq = 'C';
  18                     break;
  19             case 'g': case 'G':
  20                     *seq = 'G';
  21                     break;
  22             default:
  23             *seq = 'N';
  24                     break;
  25         }   // Whether a looking-up array[64] can be faster ?
  26         ++seq;
  27     }
  28 }
  29
  30 // *seq must have been Normalized to /[ATCGN]*/
  31 FORCE_INLINE char *ChrSeqRevComp(char const * seq, size_t len){
  32     char *revcomseq=malloc(len+1);
  33     const char *tmpseqin=seq+len-1; // if len is shorter than real length, trim it.
  34     char *tmpseqrc=revcomseq;
  35     while(tmpseqin>=seq){
  36         switch (*tmpseqin) {
  37             case 'A':
  38                     *tmpseqrc++ = 'T';
  39                     break;
  40             case 'T':
  41                     *tmpseqrc++ = 'A';
  42                     break;
  43             case 'C':
  44                     *tmpseqrc++ = 'G';
  45                     break;
  46             case 'G':
  47                     *tmpseqrc++ = 'C';
  48                     break;
  49             default:
  50             *tmpseqrc++ = 'N';
  51                     break;
  52         }   // Whether a looking-up array[32] can be faster ?
  53         --tmpseqin;
  54     }
  55     *tmpseqrc++ = '\0';
  56     return revcomseq;
  57 }
  58
  59 #endif  // chrseq.h
  60