#include <corelib/ncbistd.hpp>#include <util/range.hpp>#include <objtools/blast/seqdb_reader/seqdb.hpp>#include <algo/blast/core/blast_message.h>#include "mhfile.hpp"
Include dependency graph for blastkmerutils.hpp:
This graph shows which files directly or indirectly include this file:
Go to the source code of this file.
Go to the SVN repository for this file.
Classes | |
| struct | BlastKmerStats |
| Structure for ancillary data on KMER search. More... | |
| struct | SBlastKmerParameters |
| struct | SOneBlastKmerSearch |
Macros | |
| #define | PKMER_PRIME 1048583 |
Typedefs | |
| typedef vector< pair< uint32_t, double > > | TBlastKmerPrelimScoreVector |
| Vector of pairs of database OIDs and scores. More... | |
Functions | |
| USING_SCOPE (blast) | |
| set< uint32_t > | BlastKmerGetKmerSet (const string &query_sequence, bool do_seg, TSeqRange &range, int kmerNum, int alphabetChoice) |
| Get KMERs for a given sequence using a compressed alphabet. More... | |
| set< uint32_t > | BlastKmerGetKmerSet2 (const string &query_sequence, TSeqRange &range, int kmerNum, int alphabetChoice, vector< int > badMers) |
| Get KMERs for a given sequence using a compressed alphabet. More... | |
| set< uint32_t > | BlastKmerGetKmerSetStats (const string &query_sequence, int kmerNum, map< string, int > &kmerCount, map< string, int > &kmerCountPlus, int alphabetChoice, bool perQuery) |
| Simplified version of BlastKmerGetKmerSet. More... | |
| int | BlastKmerBreakUpSequence (int length, vector< TSeqRange > &range_v, int chunkSize) |
| Breaks a sequences up into chunks if the sequences is above a certain length. More... | |
| void | BlastKmerGetCompressedTranslationTable (vector< Uint1 > &trans_table, int alphabetChoice) |
| Creates translation table for compressed alphabets. More... | |
| int | BlastKmerGetDistance (const vector< uint32_t > &minhash1, const vector< uint32_t > &minhash2) |
| Calculates the number of differences between two minhash arrays. More... | |
| bool | minhash_query (const string &query, vector< vector< uint32_t > > &seq_hash, int num_hashes, uint32_t *a, uint32_t *b, int do_seg, int kmerNum, int alphabetChoice, int chunkSize) |
| bool | minhash_query2 (const string &query, vector< vector< uint32_t > > &seq_hash, int kmerNum, int numHashes, int alphabetChoice, vector< int > badMers, int chunkSize) |
| Hash the query for the minimum values;. More... | |
| void | get_LSH_match_from_hash (const vector< vector< uint32_t > > &lsh_hash_vec, const uint64_t *lsh_array, vector< set< uint32_t > > &candidates) |
| void | get_LSH_hashes (vector< vector< uint32_t > > &query_hash, vector< vector< uint32_t > > &lsh_hash_vec, int num_bands, int rows_per_band) |
| void | get_LSH_hashes5 (vector< vector< uint32_t > > &query_hash, vector< vector< uint32_t > > &lsh_hash_vec, int numHashes, int numRows) |
| Gets the LSH hash for one hash function. More... | |
| void | neighbor_query (const vector< vector< uint32_t > > &query_hash, const uint64_t *lsh, vector< set< uint32_t > > &candidates, CMinHashFile &mhfile, int num_hashes, int min_hits, double thresh, TBlastKmerPrelimScoreVector &score_vector, BlastKmerStats &kmer_stats, int kmerVersion) |
| void | GetRandomNumbers (uint32_t *a, uint32_t *b, int numHashes) |
| Get the random numbers for the hash function. More... | |
| void | GetKValues (vector< vector< int > > &kvector, int k_value, int l_value, int array_size) |
| Function to get the k sites to compare for Buhler LSH. More... | |
| void | get_LSH_hashes2 (vector< vector< uint32_t > > &query_hash, vector< vector< uint32_t > > &lsh_hash_vec, int num_k, int num_l, vector< vector< int > > &kValues) |
| int | BlastKmerVerifyIndex (CRef< CSeqDB > seqdb, string &error_msg) |
Macro Definition Documentation
◆ PKMER_PRIME
| #define PKMER_PRIME 1048583 |
Definition at line 48 of file blastkmerutils.hpp.
Typedef Documentation
◆ TBlastKmerPrelimScoreVector
| typedef vector< pair<uint32_t, double> > TBlastKmerPrelimScoreVector |
Vector of pairs of database OIDs and scores.
ONLY for use during KMER search, not presentation of results or communication with other modules (BLAST or not).
Definition at line 122 of file blastkmerutils.hpp.
Function Documentation
◆ BlastKmerBreakUpSequence()
Breaks a sequences up into chunks if the sequences is above a certain length.
Each chunk has an overlap with the adjoining chunks. Breaking sequences up into chunks makes the minhash procedure much more effective two sequences of different lengths are being compared.
- Parameters
-
length Total length of sequence being broken up [in] range_v Vector of ranges to be filled in [out] chunkSize number of residues in sequence chunk [in]
- Returns
- number of chunks. Should be integer more than zero. Zero or less indicates an error.
Definition at line 362 of file blastkmerutils.cpp.
References ChunkSize(), i, last(), MAX, MIN, and compile_time_bits::range().
Referenced by minhash_query(), minhash_query2(), s_MinhashSequences(), and s_MinhashSequences2().
◆ BlastKmerGetCompressedTranslationTable()
Creates translation table for compressed alphabets.
- Parameters
-
trans_table Translation table [out] alphabetChoice 0 is 15 letter, 1 is 10 letter alphabet [in]
Definition at line 330 of file blastkmerutils.cpp.
References _ASSERT, AMINOACID_TO_NCBISTDAA, i, int, isalpha(), and isspace().
Referenced by BlastKmerGetKmerSet(), BlastKmerGetKmerSet2(), and BlastKmerGetKmerSetStats().
◆ BlastKmerGetDistance()
| int BlastKmerGetDistance | ( | const vector< uint32_t > & | minhash1, |
| const vector< uint32_t > & | minhash2 | ||
| ) |
Calculates the number of differences between two minhash arrays.
Used to decide whether two arrays are similar enough. The assumption is made that both arrays are of the same size.
- Parameters
-
minhash1 First array [in] minhash2 Second array [in]
- Returns
- distance.
Definition at line 399 of file blastkmerutils.cpp.
Referenced by s_MinhashSequences(), and s_MinhashSequences2().
◆ BlastKmerGetKmerSet()
| set<uint32_t> BlastKmerGetKmerSet | ( | const string & | query_sequence, |
| bool | do_seg, | ||
| TSeqRange & | range, | ||
| int | kmerNum, | ||
| int | alphabetChoice | ||
| ) |
Get KMERs for a given sequence using a compressed alphabet.
- Parameters
-
query_sequence string with one sequence [in] do_seg Should the sequence be segged (not recommended) [in] range portion of sequence to be processed [in] kmerNum size of kmer [in] alphabetChoice 0 is 15 letter, 1 is 10 letter alphabet [in]
- Returns
- set of unsigned ints for the kmers.
Definition at line 206 of file blastkmerutils.cpp.
References BlastKmerGetCompressedTranslationTable(), BlastSeqLocFree(), free(), i, set< Key, Compare >::insert(), malloc(), BlastSeqLoc::next, NULL, query, compile_time_bits::range(), SegParametersFree(), SegParametersNewAa(), and SeqBufferSeg().
Referenced by minhash_query(), and s_MinhashSequences().
◆ BlastKmerGetKmerSet2()
| set<uint32_t> BlastKmerGetKmerSet2 | ( | const string & | query_sequence, |
| TSeqRange & | range, | ||
| int | kmerNum, | ||
| int | alphabetChoice, | ||
| vector< int > | badMers | ||
| ) |
Get KMERs for a given sequence using a compressed alphabet.
This version can read in overrepresented KMERs and extend them by one.
- Parameters
-
query_sequence string with one sequence [in] range portion of sequence to be processed [in] kmerNum size of kmer [in] alphabetChoice 0 is 15 letter, 1 is 10 letter alphabet [in] badMers Overrepresented KMERs [in]
- Returns
- set of unsigned ints for the kmers.
Definition at line 270 of file blastkmerutils.cpp.
References BlastKmerGetCompressedTranslationTable(), free(), i, set< Key, Compare >::insert(), malloc(), query, and compile_time_bits::range().
Referenced by minhash_query2(), and s_MinhashSequences2().
◆ BlastKmerGetKmerSetStats()
| set<uint32_t> BlastKmerGetKmerSetStats | ( | const string & | query_sequence, |
| int | kmerNum, | ||
| map< string, int > & | kmerCount, | ||
| map< string, int > & | kmerCountPlus, | ||
| int | alphabetChoice, | ||
| bool | perQuery | ||
| ) |
Simplified version of BlastKmerGetKmerSet.
Intended for gathering statistics on KMERS in the database.
- Parameters
-
query_sequence string with one sequence [in] kmerNum size of kmer [in]
- Returns
- param kmerCount Count population of different KMERS
- param kmerCount Count population of different KMERS one longer than kmerNum
- Parameters
-
alphabetChoice 0 is 15 letter, 1 is 10 letter alphabet [in] perQuery COunt kmers per query or total in database.
- Returns
- set of unsigned ints for the kmers.
Definition at line 132 of file blastkmerutils.cpp.
References map_checker< Container >::begin(), BlastKmerGetCompressedTranslationTable(), map_checker< Container >::end(), i, set< Key, Compare >::insert(), NStr::NumericToString(), and query.
◆ BlastKmerVerifyIndex()
Definition at line 923 of file blastkmerutils.cpp.
References CSeqDB::FindVolumePaths(), i, and s_BlastKmerVerifyVolume().
Referenced by CCkblastindexApplication::Run().
◆ get_LSH_hashes()
| void get_LSH_hashes | ( | vector< vector< uint32_t > > & | query_hash, |
| vector< vector< uint32_t > > & | lsh_hash_vec, | ||
| int | num_bands, | ||
| int | rows_per_band | ||
| ) |
Definition at line 532 of file blastkmerutils.cpp.
References b, do_pearson_hash(), ncbi::grid::netcache::search::fields::key, n, r(), and ct::sort().
Referenced by BOOST_AUTO_TEST_CASE(), and CBlastKmer::x_ProcessQuery().
◆ get_LSH_hashes2()
| void get_LSH_hashes2 | ( | vector< vector< uint32_t > > & | query_hash, |
| vector< vector< uint32_t > > & | lsh_hash_vec, | ||
| int | num_k, | ||
| int | num_l, | ||
| vector< vector< int > > & | kValues | ||
| ) |
Definition at line 651 of file blastkmerutils.cpp.
References do_pearson_hash(), i, ncbi::grid::netcache::search::fields::key, max(), n, r(), and ct::sort().
Referenced by CBlastKmer::x_ProcessQuery().
◆ get_LSH_hashes5()
| void get_LSH_hashes5 | ( | vector< vector< uint32_t > > & | query_hash, |
| vector< vector< uint32_t > > & | lsh_hash_vec, | ||
| int | numHashes, | ||
| int | numRows | ||
| ) |
Gets the LSH hash for one hash function.
- Parameters
-
query_hash Hash values for query [in] lsh_hash_vec LSH query hash [out] numHashes number of hashes in signature [in] numRows number of rows (2?) in LSH [in]
Definition at line 566 of file blastkmerutils.cpp.
References b, do_pearson_hash(), ncbi::grid::netcache::search::fields::key, n, r(), and ct::sort().
Referenced by BOOST_AUTO_TEST_CASE(), and CBlastKmer::x_ProcessQuery().
◆ get_LSH_match_from_hash()
| void get_LSH_match_from_hash | ( | const vector< vector< uint32_t > > & | lsh_hash_vec, |
| const uint64_t * | lsh_array, | ||
| vector< set< uint32_t > > & | candidates | ||
| ) |
Definition at line 686 of file blastkmerutils.cpp.
References i.
Referenced by CBlastKmer::x_RunKmerFile().
◆ GetKValues()
Function to get the k sites to compare for Buhler LSH.
Definition at line 629 of file blastkmerutils.cpp.
References CRandom::GetRand(), and i.
Referenced by CBlastKmerBuildIndex::x_BuildIndex().
◆ GetRandomNumbers()
Get the random numbers for the hash function.
Definition at line 613 of file blastkmerutils.cpp.
References a, b, CRandom::GetRand(), i, and PKMER_PRIME.
Referenced by CBlastKmerBuildIndex::x_BuildIndex().
◆ minhash_query()
| bool minhash_query | ( | const string & | query, |
| vector< vector< uint32_t > > & | seq_hash, | ||
| int | num_hashes, | ||
| uint32_t * | a, | ||
| uint32_t * | b, | ||
| int | do_seg, | ||
| int | kmerNum, | ||
| int | alphabetChoice, | ||
| int | chunkSize | ||
| ) |
Definition at line 415 of file blastkmerutils.cpp.
References a, b, set< Key, Compare >::begin(), BlastKmerBreakUpSequence(), BlastKmerGetKmerSet(), set< Key, Compare >::empty(), set< Key, Compare >::end(), i, query, and uhash().
Referenced by BOOST_AUTO_TEST_CASE(), and CBlastKmer::x_ProcessQuery().
◆ minhash_query2()
| bool minhash_query2 | ( | const string & | query, |
| vector< vector< uint32_t > > & | seq_hash, | ||
| int | kmerNum, | ||
| int | numHashes, | ||
| int | alphabetChoice, | ||
| vector< int > | badMers, | ||
| int | chunkSize | ||
| ) |
Hash the query for the minimum values;.
- Parameters
-
query as a ASCII string [in] seq_hash hash values for all kmers [out] kmerNum number of letters in a KMER [in] numHashes number of hashes in a signature [in] alphabetChoice 15 or 10 letters [in] badMers Overrepresented KMERS [in]
Definition at line 479 of file blastkmerutils.cpp.
References set< Key, Compare >::begin(), BlastKmerBreakUpSequence(), BlastKmerGetKmerSet2(), set< Key, Compare >::empty(), set< Key, Compare >::end(), FNV_hash(), i, query, and ct::sort().
Referenced by BOOST_AUTO_TEST_CASE(), and CBlastKmer::x_ProcessQuery().
◆ neighbor_query()
| void neighbor_query | ( | const vector< vector< uint32_t > > & | query_hash, |
| const uint64_t * | lsh, | ||
| vector< set< uint32_t > > & | candidates, | ||
| CMinHashFile & | mhfile, | ||
| int | num_hashes, | ||
| int | min_hits, | ||
| double | thresh, | ||
| TBlastKmerPrelimScoreVector & | score_vector, | ||
| BlastKmerStats & | kmer_stats, | ||
| int | kmerVersion | ||
| ) |
Definition at line 744 of file blastkmerutils.cpp.
References map_checker< Container >::begin(), map_checker< Container >::end(), estimate_jaccard(), estimate_jaccard2(), CMinHashFile::GetDataWidth(), CMinHashFile::GetHits(), CMinHashFile::GetMinHits(), CMinHashFile::GetVersion(), BlastKmerStats::hit_count, i, map_checker< Container >::insert(), int, BlastKmerStats::jd_count, BlastKmerStats::jd_oid_count, n, offset, BlastKmerStats::oids_considered, read_size(), s_HashHashQuery(), ct::sort(), and BlastKmerStats::total_matches.
Referenced by CBlastKmer::x_RunKmerFile().
◆ USING_SCOPE()
| USING_SCOPE | ( | blast | ) |
