update headers, use const for better performance (.500ms on ~2gb file), update comments for functions

1 files changed, 14 insertions, 15 deletions

diff --git a/kmer_utils.c b/kmer_utils.c
index 7b483ea..89df3b2 100644
--- a/kmer_utils.c
+++ b/kmer_utils.c
@@ -4,19 +4,9 @@
 #include <string.h>
 #include <unistd.h>
 
-// This function takes a char array containing sequences and converts it
-// into a kmer index (essentially a base 4 radix converstion to base 10, with
-// some mumbo-jumbo of taking each of the characters we want (ACGT) and turning
-// them into a radix-4 number we can use.
-//
-// kind of convoluted but it works.
-//
-// Arguemnts: 
-//  char *str - a NULL terminated character array 
-//  long kmer - how long of a index value you want to return
-//  long error_pos - what index to return for a non ACGT character
-//
-inline long num_to_index(const char *str, int kmer, long error_pos) {
+// convert a string of k-mer size base-4 values  into a
+// base-10 index
+long num_to_index(const char *str, const int kmer, const long error_pos) {
 
   int i = 0;
   unsigned long out = 0;
@@ -24,6 +14,9 @@ inline long num_to_index(const char *str, int kmer, long error_pos) {
 
   for(i = kmer - 1; i >= 0; i--){
 
+    if(str[i] >> 2)
+      return error_pos;
+
     out += str[i] * multiply;
     multiply = multiply << 2;
   }
@@ -31,11 +24,15 @@ inline long num_to_index(const char *str, int kmer, long error_pos) {
   return out;
 }
 
-void convert_kmer_to_num(char *str, long length) {
+// replaces values in a char array of ACGT's and others with
+// values that correspond to their base 4 value to be used in
+// num_to_index.
+void convert_kmer_to_num(char *str, const long length) {
 
   long i = 0;
 
   for(i = 0; i < length; i++) {
+    // this is _not_ portable, only works with ASCII values.
     switch(str[i] | 0x20 ) {
       case 'a':
         str[i] = 0;
@@ -50,7 +47,9 @@ void convert_kmer_to_num(char *str, long length) {
         str[i] = 3;
         break;
       default:
-        str[i] = 5;
+        // Error Checking: use 4 so we can shift right twice
+        // to check quickly is there is an non ACGT character 
+        str[i] = 4;
     }
 
   }