v. 2.3.0

short k-mers (k<10) support
refresh-bio · Sep 10, 2015 · 0635dae · 0635dae
1 parent 78601be
commit 0635dae
Show file tree

Hide file tree

Showing 112 changed files with 13,063 additions and 261 deletions.
diff --git a/API.pdf b/API.pdf
diff --git a/kmc_api/kmc_file.cpp b/kmc_api/kmc_file.cpp
diff --git a/kmc_api/kmc_file.h b/kmc_api/kmc_file.h
@@ -4,9 +4,9 @@
 
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
 
-  Version: 2.2.0
-  Date   : 2015-04-15
-*/
+  Version: 2.3.0
+  Date   : 2015-08-21
+ */
 
 #ifndef _KMC_FILE_H
 #define _KMC_FILE_H
@@ -16,6 +16,19 @@
 #include <string>
 #include <vector>
 
+struct CKMCFileInfo
+{
+	uint32 kmer_length;
+	uint32 mode;
+	uint32 counter_size;
+	uint32 lut_prefix_length;
+	uint32 signature_len;	
+	uint32 min_count;
+	uint64 max_count;
+	bool both_strands;
+	uint64 total_kmers;
+};
+
 class CKMCFile
 {
 	enum open_mode {closed, opened_for_RA, opened_for_listing};
@@ -44,19 +57,20 @@ class CKMCFile
 	uint32 lut_prefix_length;
 	uint32 signature_len;
 	uint32 min_count;
-	uint32 max_count;
+	uint64 max_count;
 	uint64 total_kmers;
+	bool both_strands;
 
 	uint32 kmc_version;
 	uint32 sufix_size;		// sufix's size in bytes 
 	uint32 sufix_rec_size;  // sufix_size + counter_size
 
 	uint32 original_min_count;
-	uint32 original_max_count;
+	uint64 original_max_count;
 
 	static uint64 part_size; // the size of a block readed to sufix_file_buf, in listing mode 
 
-	bool BinarySearch(int64 index_start, int64 index_stop, const CKmerAPI& kmer, uint32& counter, uint32 pattern_offset);
+	bool BinarySearch(int64 index_start, int64 index_stop, const CKmerAPI& kmer, uint64& counter, uint32 pattern_offset);
 
 	// Open a file, recognize its size and check its marker. Auxiliary function.
 	bool OpenASingleFile(const std::string &file_name, FILE *&file_handler, uint64 &size, char marker[]);	
@@ -67,8 +81,17 @@ class CKMCFile
 	// Reload a contents of an array "sufix_file_buf" for listing mode. Auxiliary function. 
 	void Reload_sufix_file_buf();
 
-	// Implementation of GetCountersForRead for kmc1 database format
-	bool GetCountersForRead_kmc1(const std::string& read, std::vector<uint32>& counters);
+	// Implementation of GetCountersForRead for kmc1 database format for both strands
+	bool GetCountersForRead_kmc1_both_strands(const std::string& read, std::vector<uint32>& counters);
+
+	// Implementation of GetCountersForRead for kmc1 database format without choosing canonical k-mer
+	bool GetCountersForRead_kmc1(const std::string& read, std::vector<uint32>& counters);		
+
+	using super_kmers_t = std::vector<std::tuple<uint32, uint32, uint32>>;//start_pos, len, bin_no
+	void GetSuperKmers(const std::string& transformed_read, super_kmers_t& super_kmers);
+
+	// Implementation of GetCountersForRead for kmc2 database format for both strands
+	bool GetCountersForRead_kmc2_both_strands(const std::string& read, std::vector<uint32>& counters);
 
 	// Implementation of GetCountersForRead for kmc2 database format
 	bool GetCountersForRead_kmc2(const std::string& read, std::vector<uint32>& counters);
@@ -86,6 +109,8 @@ class CKMCFile
 	// Return next kmer in CKmerAPI &kmer. Return its counter in float &count. Return true if not EOF
 	bool ReadNextKmer(CKmerAPI &kmer, float &count);
 
+	bool ReadNextKmer(CKmerAPI &kmer, uint64 &count); //for small k-values when counter may be longer than 4bytes
+
 	bool ReadNextKmer(CKmerAPI &kmer, uint32 &count);
 	// Release memory and close files in case they were opened 
 	bool Close();
@@ -100,7 +125,10 @@ class CKMCFile
 	bool SetMaxCount(uint32 x);
 
 	// Return a value of max_count. Kmers with counters above this theshold are ignored 
-	uint32 GetMaxCount(void);
+	uint64 GetMaxCount(void);
+
+	//Return true if kmc was run without -b switch.
+	bool GetBothStrands(void);
 
 	// Return the total number of kmers between min_count and max_count
 	uint64 KmerCount(void);
@@ -119,15 +147,20 @@ class CKMCFile
 
 	bool CheckKmer(CKmerAPI &kmer, uint32 &count);
 
+	bool CheckKmer(CKmerAPI &kmer, uint64 &count);
+
 	// Return true if kmer exists
 	bool IsKmer(CKmerAPI &kmer);
 
 	// Set original (readed from *.kmer_pre) values for min_count and max_count
 	void ResetMinMaxCounts(void);
 
 	// Get current parameters from kmer_database
-	bool Info(uint32 &_kmer_length, uint32 &_mode, uint32 &_counter_size, uint32 &_lut_prefix_length, uint32 &_signature_len, uint32 &_min_count, uint32 &_max_count, uint64 &_total_kmers);
+	bool Info(uint32 &_kmer_length, uint32 &_mode, uint32 &_counter_size, uint32 &_lut_prefix_length, uint32 &_signature_len, uint32 &_min_count, uint64 &_max_count, uint64 &_total_kmers);
 
+	// Get current parameters from kmer_database
+	bool Info(CKMCFileInfo& info);
+
 	// Get counters for all k-mers in read
 	bool GetCountersForRead(const std::string& read, std::vector<uint32>& counters);
 	bool GetCountersForRead(const std::string& read, std::vector<float>& counters);

diff --git a/kmc_api/kmer_api.cpp b/kmc_api/kmer_api.cpp
@@ -4,8 +4,8 @@
 
   Authors: Sebastian Deorowicz and Agnieszka Debudaj-Grabysz
 
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 

diff --git a/kmc_api/kmer_api.h b/kmc_api/kmer_api.h
@@ -4,8 +4,8 @@ The homepage of the KMC project is http://sun.aei.polsl.pl/kmc
 
 Authors: Sebastian Deorowicz and Agnieszka Debudaj-Grabysz
 
-Version: 2.2.0
-Date   : 2015-04-15
+Version: 2.3.0
+Date   : 2015-08-21
 */
 
 #ifndef _KMER_API_H
@@ -15,6 +15,7 @@ Date   : 2015-04-15
 #include "kmer_defs.h"
 #include <string>
 #include <iostream>
+#include <vector>
 #include "mmer.h"
 class CKMCFile;
 
@@ -65,6 +66,20 @@ class CKmerAPI
 		}
 		kmer_data[no_of_rows - 1] += (uint64)val << (62 - (((kmer_length - 1 + byte_alignment) & 31) * 2));
 	}
+
+	//----------------------------------------------------------------------------------
+	inline void SHR_insert2bits(uchar val)
+	{
+		for (uint32 i = no_of_rows - 1; i > 0; --i)
+		{
+			kmer_data[i] >>= 2;
+			kmer_data[i] += kmer_data[i - 1] << 62;
+		}
+		kmer_data[0] >>= 2;
+		kmer_data[no_of_rows - 1] &= ~((1ull << ((32 - (kmer_length + byte_alignment - (no_of_rows - 1) * 32)) * 2)) - 1);//mask falling of symbol
+		kmer_data[0] += ((uint64)val << 62) >> (byte_alignment * 2);
+	}
+
 	// ----------------------------------------------------------------------------------
 	inline void from_binary(const char* kmer)
 	{
@@ -73,6 +88,14 @@ class CKmerAPI
 			insert2bits(i, kmer[i]);
 	}
 
+	// ----------------------------------------------------------------------------------
+	inline void from_binary_rev(const char* kmer)
+	{
+		clear();
+		for (uint32 i = 0; i < kmer_length; ++i)
+			insert2bits(i, 3 - kmer[kmer_length - i - 1]);
+	}
+
 	// ----------------------------------------------------------------------------------
 	template<typename RandomAccessIterator>
 	inline void to_string_impl(RandomAccessIterator iter)
@@ -424,6 +447,19 @@ class CKmerAPI
 		str[kmer_length] = '\0';
 	};
 
+
+	inline void to_long(std::vector<uint64>& kmer)
+	{
+		kmer.resize(no_of_rows);
+		uint32 offset = 62 - ((kmer_length - 1 + byte_alignment) & 31) * 2;
+		for (int32 i = no_of_rows - 1; i >= 1; --i)
+		{
+			kmer[i] = kmer_data[i] >> offset;
+			kmer[i] += kmer_data[i - 1] << (64 - offset);
+		}
+		kmer[0] = kmer_data[0] >> offset;
+	}
+
 	//-----------------------------------------------------------------------
 	// Convert kmer into string (an alphabet ACGT)
 	// OUT 	: str - string kmer

diff --git a/kmc_api/kmer_defs.h b/kmc_api/kmer_defs.h
@@ -4,16 +4,16 @@
 
   Authors: Sebastian Deorowicz and Agnieszka Debudaj-Grabysz
 
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 
 #ifndef _KMER_DEFS_H
 #define _KMER_DEFS_H
 
-#define KMC_VER		"2.2.0"
-#define KMC_DATE	"2015-04-15"
+#define KMC_VER		"2.3.0"
+#define KMC_DATE	"2015-08-21"
 
 #define MIN(x,y)	((x) < (y) ? (x) : (y))
 

diff --git a/kmc_api/mmer.cpp b/kmc_api/mmer.cpp
@@ -5,8 +5,8 @@
   
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
   
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 #include "../kmc_api/mmer.h"

diff --git a/kmc_api/mmer.h b/kmc_api/mmer.h
@@ -4,8 +4,8 @@
   
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
   
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 #ifndef _MMER_H

diff --git a/kmc_dump/kmc_dump.cpp b/kmc_dump/kmc_dump.cpp
@@ -7,8 +7,8 @@
 
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
 
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 #include "stdafx.h"
@@ -85,7 +85,7 @@ int _tmain(int argc, char* argv[])
 		uint32 _lut_prefix_length;
 		uint32 _signature_len;
 		uint32 _min_count;
-		uint32 _max_count;
+		uint64 _max_count;
 		uint64 _total_kmers;
 
 		kmer_data_base.Info(_kmer_length, _mode, _counter_size, _lut_prefix_length, _signature_len, _min_count, _max_count, _total_kmers);
@@ -118,7 +118,7 @@ int _tmain(int argc, char* argv[])
 		}
 		else
 		{
-			uint32 counter;
+			uint64 counter;
 			while (kmer_data_base.ReadNextKmer(kmer_object, counter))
 			{
 				kmer_object.to_string(str);

diff --git a/kmc_dump/nc_utils.cpp b/kmc_dump/nc_utils.cpp
@@ -7,8 +7,8 @@
 
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
 
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 #include "stdafx.h"

diff --git a/kmc_dump/nc_utils.h b/kmc_dump/nc_utils.h
@@ -7,8 +7,8 @@
 
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
 
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 #include <string>

diff --git a/kmc_dump_sample/kmc_dump_sample.cpp b/kmc_dump_sample/kmc_dump_sample.cpp
@@ -7,8 +7,8 @@
 
   Authors: Sebastian Deorowicz, Agnieszka Debudaj-Grabysz, Marek Kokot
 
-  Version: 2.2.0
-  Date   : 2015-04-15
+  Version: 2.3.0
+  Date   : 2015-08-21
 */
 
 #include "stdafx.h"
@@ -83,7 +83,7 @@ int _tmain(int argc, char* argv[])
 		uint32 _lut_prefix_length;
 		uint32 _signature_len;
 		uint32 _min_count;
-		uint32 _max_count;
+		uint64 _max_count;
 		uint64 _total_kmers;
 
 		kmer_data_base.Info(_kmer_length, _mode, _counter_size, _lut_prefix_length, _signature_len, _min_count, _max_count, _total_kmers);

diff --git a/kmc_tools.pdf b/kmc_tools.pdf
diff --git a/kmc_tools/asmlib_wrapper.h b/kmc_tools/asmlib_wrapper.h
@@ -0,0 +1,23 @@
+/*
+  This file is a part of KMC software distributed under GNU GPL 3 licence.
+  The homepage of the KMC project is http://sun.aei.polsl.pl/kmc
+  
+  Authors: Marek Kokot
+  
+  Version: 2.3.0
+  Date   : 2015-08-21
+*/
+
+
+#ifndef _ASMLIB_WRAPPER_H
+#define _ASMLIB_WRAPPER_H
+
+#include "defs.h"
+#ifdef DISABLE_ASMLIB
+#define A_memcpy memcpy
+#define SetMemcpyCacheLimit(X) 
+#else
+#include "libs/asmlib.h"
+#endif
+
+#endif