minor changes to hash function, inconsequential to SBT operation

medvedevgroup · Jul 20, 2019 · 3edcb49 · 3edcb49
1 parent 9f3b8a0
commit 3edcb49
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 0 deletions.
diff --git a/cmd_sabuhash_test.cc b/cmd_sabuhash_test.cc
@@ -49,6 +49,7 @@ void SabuhashTestCommand::usage
 	s << "  --strings        hash strings instead of bits; applies only when input is" << endl;
 	s << "                   from sequence files" << endl;
 	s << "                   (by default kmers are hashed in 2-bit encoded form)" << endl;
+	s << "  --negate         negate hash values; replace values by their ones-complement" << endl;
 	s << "  --modulus=<M>    set the hash modulus" << endl;
 	s << "                   (by default, the hash values have no modulus)" << endl;
 	s << "  --seed=<number>  set the hash function's 32-bit seed" << endl;
@@ -75,6 +76,7 @@ void SabuhashTestCommand::parse
 
 	kmerSize       = defaultKmerSize;
 	useStringKmers = false;
+	negateHash     = false;
 	modulus        = 0;
 	hashSeed       = 0;
 #ifdef useJellyHash
@@ -136,6 +138,11 @@ void SabuhashTestCommand::parse
 		if (arg == "--strings")
 		 	{ useStringKmers = true;  continue; }
 
+		// --negate
+
+		if (arg == "--negate")
+		 	{ negateHash = true;  continue; }
+
 		// --modulus=<M>
 
 		if ((is_prefix_of (arg, "--modulus="))
@@ -278,6 +285,13 @@ void SabuhashTestCommand::perform_hash_test
 		h64  = hasher->hash(seq);
 		h64r = hasher->hash(revCompSeq);
 		}
+
+	if (negateHash)
+		{
+		h64  = ~h64;
+		h64r = ~h64r;
+		}
+
 	if (modulus == 0)
 		isMatch = (h64 == h64r);
 	else

diff --git a/cmd_sabuhash_test.h b/cmd_sabuhash_test.h
@@ -28,6 +28,7 @@ class SabuhashTestCommand: public Command
 	std::vector<std::string> seqFilenames;
 	std::uint32_t kmerSize;
 	bool useStringKmers;
+	bool negateHash;
 	std::uint32_t modulus;
 	std::uint64_t hashSeed;
 	HashCanonical* hasher = nullptr;

diff --git a/sabuhash.h b/sabuhash.h
@@ -129,6 +129,7 @@ class SabuHash
 	unsigned int  k;
 	std::uint64_t seed;
 	bool          allowN;
+	bool          validNt;  // true => latest nucleotide was valid
 	unsigned int  charsAccumulated;
 	std::uint64_t kernelTable[256], forwardByK[256], forwardByK2[4];
 	std::uint64_t hForward;
@@ -141,6 +142,7 @@ class SabuHash
 	  :	k(_k),
 	    seed(_seed),
 	    allowN(_allowN),
+	    validNt(false),
 	    charsAccumulated(0),
 	    hForward(0)
 		{
@@ -258,11 +260,13 @@ class SabuHash
 		if (kernelTable[chIn] == 0)
 			{
 			// chIn not in {A,C,G,T}
+			validNt = false;
 		    charsAccumulated = 0;
 			hForward = 0;
 			return 0;
 			}
 
+		validNt = true;
 		if (charsAccumulated < k)
 			{
 			hForward = forward(hForward) ^ kernelTable[chIn];
@@ -371,6 +375,7 @@ class SabuHashCanonical
 	unsigned int  k;
 	std::uint64_t seed;
 	bool          allowN;
+	bool          validNt;  // true => latest nucleotide was valid
 	unsigned int  charsAccumulated;
 	std::uint64_t kernelTable[256], kernelTable_RC[256], kernelTable2_RC[4];
 	std::uint64_t forwardByK[256],  forwardByK_RC[256],  forwardByK2[4];
@@ -384,6 +389,7 @@ class SabuHashCanonical
 	  :	k(_k),
 	    seed(_seed),
 	    allowN(_allowN),
+	    validNt(false),
 	    charsAccumulated(0),
 	    hForward(0),
 	    hRevComp(0)
@@ -545,11 +551,13 @@ class SabuHashCanonical
 		if (kernelTable[chIn] == 0)
 			{
 			// chIn not in {A,C,G,T}
+			validNt = false;
 		    charsAccumulated = 0;
 			hForward = hRevComp = 0;
 			return 0;
 			}
 
+		validNt = true;
 		if (charsAccumulated < k)
 			{
 			hForward = SabuHash::forward(hForward) ^ kernelTable[chIn];