Merge pull request #21 from jlab/restructure_rna

restructure rna.hh and rnalib.*

librna/rnalib.c

@@ -70,32 +70,27 @@ void librna_read_param_file(const char *filename)
      x = 5' base of the basepair
      y = 3' base of the basepair
 */
-static int bp_index(char x, char y)
+int bp_index(char x, char y)
 {
   switch (x) {
     case A_BASE : switch (y) {
         case U_BASE : return AU_BP;
-        case GAP_BASE : return NO_BP;
       }
       break;
     case C_BASE : switch (y) {
         case G_BASE : return CG_BP;
-        case GAP_BASE : return NO_BP;
       }
       break;
     case G_BASE : switch (y) {
         case C_BASE : return GC_BP;
         case U_BASE : return GU_BP;
-        case GAP_BASE : return NO_BP;
       }
       break;
     case U_BASE : switch (y) {
         case G_BASE : return UG_BP;
         case A_BASE : return UA_BP;
-        case GAP_BASE : return NO_BP;
       }
       break;
-    case GAP_BASE : return NO_BP;
   }
   return NO_BP;
 }
@@ -199,16 +194,7 @@ static void decode(char *s, const char *x, const int len)
 {
 	unsigned int i;
 	for (i = 0; i < len; ++i) {
-	    switch (x[i]) {
-	      case 0 : s[i] = 'N'; break;
-	      case 1 : s[i] = 'A'; break;
-	      case 2 : s[i] = 'C'; break;
-	      case 3 : s[i] = 'G'; break;
-	      case 4 : s[i] = 'U'; break;
-	      case 5 : s[i] = '_'; break;
-		  case 6 : s[i] = '+'; break;
-	      default: abort();
-	    };
+    s[i] = BASE_CHARS[x[i]];
 	}
 }
 

librna/rnalib.h

@@ -7,7 +7,11 @@
 #include <stdbool.h>
 #endif
 
-enum base_t { N_BASE, A_BASE, C_BASE, G_BASE, U_BASE, GAP_BASE, SEPARATOR_BASE };
+enum base_t {
+  N_BASE, A_BASE, C_BASE, G_BASE, U_BASE, GAP_BASE, SEPARATOR_BASE };
+static char BASE_CHARS[SEPARATOR_BASE+1] = {
+ 'N',    'A',    'C',    'G',    'U',    '_',      '+'};
+
 enum iupac_t { N_IUPAC = 0,
   B_IUPAC = 7,
   D_IUPAC = 8,
@@ -57,5 +61,6 @@ double scale(int x);
 
 
 bool iupac_match(char base, char iupac_base);
+int bp_index(char x, char y);
 
 #endif

rtlib/rna.hh

@@ -40,34 +40,13 @@ inline bool basepairing(const alphabet *seq,
 {
   if (j<=i+1)
     return false;
-  char a = seq[i];
-  char b = seq[j-1];
 
-  switch (a) {
-    case A_BASE :
-      switch (b) {
-        case U_BASE : return true;
-      }
-      break;
-    case U_BASE :
-      switch (b) {
-        case A_BASE : return true;
-        case G_BASE : return true;
-      }
-      break;
-    case G_BASE :
-      switch (b) {
-        case C_BASE : return true;
-        case U_BASE : return true;
-      }
-      break;
-    case C_BASE :
-      switch (b) {
-        case G_BASE : return true;
-      }
-      break;
+  int basepair = bp_index(seq[i], seq[j-1]);
+  if ((basepair == N_BP) || (basepair == NO_BP)) {
+    return false;
+  } else {
+    return true;
   }
-  return false;
 }
 
 template<typename alphabet, typename pos_type, typename T>
@@ -145,34 +124,25 @@ class BaseException : public std::exception {
     }
 };
 
-inline char char_to_base(char a)
+inline int char_to_base(char a)
 {
-  char c = lower_case(a);
-  switch (c) {
-    case 'n' : return N_BASE;
-    case 'a' : return A_BASE;
-    case 'c' : return C_BASE;
-    case 'g' : return G_BASE;
-    case 'u' : return U_BASE;
-    case '_' : return GAP_BASE;
-    case '+' : return SEPARATOR_BASE;
-    default: throw BaseException(a);
-  };
+  char c = upper_case(a);
+  for (int i = 0; i <= SEPARATOR_BASE; ++i) {
+    if (c == BASE_CHARS[i]) {
+      return i;
+    }
+  }
+  throw BaseException(a);
   return 0;
 }
 
-inline char base_to_char(char a)
+inline char base_to_char(int a)
 {
-  switch (a) {
-    case N_BASE : return 'n';
-    case A_BASE : return 'a';
-    case C_BASE : return 'c';
-    case G_BASE : return 'g';
-    case U_BASE : return 'u';
-    case GAP_BASE : return '_';
-    case SEPARATOR_BASE : return '+';
-    default: throw BaseException(a);
-  };
+  if (a < SEPARATOR_BASE+1) {
+    return BASE_CHARS[a];
+  } else {
+    throw BaseException(a);
+  }
   return 0;
 }
 
@@ -226,7 +196,7 @@ inline void append_deep_rna(rope::Ref<X> &str, const Basic_Subsequence<alphabet,
 {
   for (typename Basic_Subsequence<alphabet, pos_type>::const_iterator i = sub.begin();
       i != sub.end(); ++i)
-    str.append(base_to_char(*i));
+    str.append((char) base_to_char(*i));
 }
 
 // ======== energy wrapper function ========
@@ -275,7 +245,7 @@ template<typename alphabet, typename pos_type>
 inline int hl_energy_stem(const Basic_Subsequence<alphabet, pos_type> &a)
 {
   int energy = 0;
-  
+
   for (unsigned k = 0; k < a.seq->rows(); k++)
     energy += hl_energy_stem(a.seq->row(k), a.i-1, a.j);
 

rtlib/sequence.hh

@@ -287,7 +287,7 @@ class Basic_Sequence {
       return *this;
     }
 
-    alphabet &operator[](pos_type x) 
+    alphabet &operator[](pos_type x)
     {
       assert(x < n);
       return seq[x];
@@ -328,13 +328,25 @@ inline char lower_case(char c)
   if (c == '_')
     return c;
   if (c == '+')
-	return c;
+	  return c;
   if (c < 'a')
     return c+('a'-'A');
   else
     return c;
 }
 
+inline char upper_case(char c)
+{
+  if (c == '_')
+    return c;
+  if (c == '+')
+	  return c;
+  if (c >= 'a')
+    return c-('a'-'A');
+  else
+    return c;
+}
+
 //template<typename alphabet, typename pos_type>
 inline const char &column(const M_Char /*<alphabet, pos_type>*/ &c, /*pos_type*/ unsigned l)
 {

testdata/unittest/rna.cc

@@ -46,7 +46,7 @@ BOOST_AUTO_TEST_CASE ( app_subseq_rna )
   Subsequence x(s, 3, 6);
   append_deep_rna(r, x);
   Rope q;
-  append(q, "uag");
+  append(q, "UAG");
   CHECK_EQ(r, q);
 }
 
@@ -88,15 +88,15 @@ BOOST_AUTO_TEST_CASE ( iupac )
   CHECK(!filter.query(6, 14));
   CHECK(filter.query(6, 22));
   CHECK(filter.query(16, 21));
-	
+
   char sequence2[14] = "ccacuccucccgg";
   Sequence seq2(sequence2);
   char_to_rna(seq2);
   iupac_filter<char, unsigned> filter2;
   const char pattern2[] = "ccuccuccc";
   filter2.init(seq2,pattern2);
   CHECK(!filter2.query(2,11));
-  
+
   iupac_filter<char, unsigned> filter3;
   const char pattern3[] = "acuccuccc";
   filter3.init(seq2,pattern3);