vector < vector <uint32_t> >& seq_hash,
240vector < vector <uint32_t> >& seq_hash,
254vector < vector <uint32_t> >& lsh_hash_vec,
265vector < vector <uint32_t> >&lsh_hash_vec,
270 void neighbor_query(
constvector < vector <uint32_t> >& query_hash,
289 void GetKValues(vector< vector <int> >& kvector,
300vector < vector <uint32_t> >&lsh_hash_vec,
303vector< vector<int> >& kValues);
#define NCBI_XBLAST_EXPORT
NULL operations for other cases.
Structures for BLAST messages.
EBlastSeverity
Blast error message severities .
vector< pair< uint32_t, double > > TBlastKmerPrelimScoreVector
Vector of pairs of database OIDs and scores.
void BlastKmerGetCompressedTranslationTable(vector< Uint1 > &trans_table, int alphabetChoice)
Creates translation table for compressed alphabets.
int BlastKmerVerifyIndex(CRef< CSeqDB > seqdb, string &error_msg)
int BlastKmerGetDistance(const vector< uint32_t > &minhash1, const vector< uint32_t > &minhash2)
Calculates the number of differences between two minhash arrays.
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 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.
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.
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.
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;.
void GetRandomNumbers(uint32_t *a, uint32_t *b, int numHashes)
Get the random numbers for the hash function.
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.
void get_LSH_match_from_hash(const vector< vector< uint32_t > > &lsh_hash_vec, const uint64_t *lsh_array, vector< set< uint32_t > > &candidates)
int BlastKmerBreakUpSequence(int length, vector< TSeqRange > &range_v, int chunkSize)
Breaks a sequences up into chunks if the sequences is above a certain length.
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)
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)
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.
Access data in Minhash files.
Include a standard set of the NCBI C++ Toolkit most basic headers.
#define END_NCBI_SCOPE
End previously defined NCBI scope.
#define BEGIN_NCBI_SCOPE
Define ncbi namespace.
Defines BLAST database access classes.
Structure for ancillary data on KMER search.
int jd_count
How often was the Jaccard distance calculated.
int total_matches
How many matches returned.
int num_sequences
Number of database sequences considered (in this volume)
int oids_considered
How many OIDs were considered as candidates.
int hit_count
How many hits to the hash array were there?
BlastKmerStats(int hit_count=0, int jd_count=0, int jd_oid_count=0, int oids_considered=0, int total_matches=0, int seqs=0)
Constructor.
int jd_oid_count
How many OIDs was the Jaccard distance calculated for.
int version
Version of index used (0 indicates default).
int chunkSize
size of a query chunk to process (default is 150).
int numHashes
Number of hash functions per signature.
int samples
Number of samples of query signature are made?
int rowsPerBand
Number of values sampled from signature.
SBlastKmerParameters(int numHashes, int rowsPerBand, int samples, int kmerNum, int alphabetChoice, int version, int chunkSize=150)
Constructor.
int alphabetChoice
15 or 10 letter alphabet (0 for 15, 1 for 10).
int kmerNum
number of letters in KMER.
vector< TBlastKmerPrelimScoreVector > scoreVector
Scores for one query.
EBlastSeverity severity
Error or warning (only use if status is non-zero).
int status
Status of the query (0 is good, otherwise an error has occurred)
vector< vector< uint32_t > > queryLSHHash
LSH Hashes for one query (multiple chunks)
vector< BlastKmerStats > kmerStatsVector
Stats for one query.
SOneBlastKmerSearch(int numFiles=0)
Constructor.
vector< vector< uint32_t > > queryHash
Hashes for one query (multiple chunks)
CRef< CSeq_id > qSeqid
Seqid of the query.
string errDescription
Error description.
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4