src/algo/blast/core/blast_util.c File Reference

Detailed Description

Various BLAST utilities.

Definition in file blast_util.c.

#include <algo/blast/core/blast_util.h>
#include <algo/blast/core/blast_filter.h>
#include <algo/blast/core/blast_stat.h>

Include dependency graph for blast_util.c:

Go to the source code of this file.

Functions

void __sfree (void **x)
 Implemented in blast_util.c.
SSeqRange SSeqRangeNew (Int4 start, Int4 stop)
 Create a new SSeqRange structure with both fields initialized.
Int4 SSeqRangeArrayLessThanOrEqual (const SSeqRange *ranges, Int4 num_ranges, Int4 target)
 Returns the index of the range, such that this element is the first range that either contains the target or if no such range exists, the index of the first range, such that the target is less than this range.
static void s_BlastSequenceBlkFreeSeqRanges (BLAST_SequenceBlk *seq_blk)
 Auxiliary function to free the BLAST_SequenceBlk::seq_ranges field if applicable.
Int2 BlastSetUp_SeqBlkNew (const Uint1 *buffer, Int4 length, BLAST_SequenceBlk **seq_blk, Boolean buffer_allocated)
 Allocates memory for *sequence_blk and then populates it.
Int2 BlastSeqBlkNew (BLAST_SequenceBlk **retval)
 Allocates a new sequence block structure.
Int2 BlastSeqBlkSetSequence (BLAST_SequenceBlk *seq_blk, const Uint1 *sequence, Int4 seqlen)
 Stores the sequence in the sequence block structure.
Int2 BlastSeqBlkSetCompressedSequence (BLAST_SequenceBlk *seq_blk, const Uint1 *sequence)
 Stores the compressed nucleotide sequence in the sequence block structure for the subject sequence when BLASTing 2 sequences.
Int2 BlastSeqBlkSetSeqRanges (BLAST_SequenceBlk *seq_blk, SSeqRange *seq_ranges, Uint4 num_seq_ranges, Boolean copy_seq_ranges)
 Sets the seq_range and related fields appropriately in the BLAST_SequenceBlk structure.
void BlastSequenceBlkClean (BLAST_SequenceBlk *seq_blk)
 Deallocate memory only for the sequence in the sequence block.
BLAST_SequenceBlk * BlastSequenceBlkFree (BLAST_SequenceBlk *seq_blk)
 Deallocate memory for a sequence block.
void BlastSequenceBlkCopy (BLAST_SequenceBlk **copy, BLAST_SequenceBlk *src)
 Copies contents of the source sequence block without copying sequence buffers; sets all "field_allocated" booleans to FALSE, to make sure fields are not freed on the call to BlastSequenceBlkFree.
Int2 BlastProgram2Number (const char *program, EBlastProgramType *number)
 Set number for a given program type.
Int2 BlastNumber2Program (EBlastProgramType number, char **program)
 Return string name for program given a number.
static Uint1 s_CodonToAA (Uint1 *codon, const Uint1 *codes)
 Translate 3 nucleotides into an amino acid MUST have 'X' as unknown amino acid.
Int4 BLAST_GetTranslation (const Uint1 *query_seq, const Uint1 *query_seq_rev, Int4 nt_length, Int2 frame, Uint1 *prot_seq, const Uint1 *genetic_code)
 GetTranslation to get the translation of the nucl.
Int2 BlastCompressBlastnaSequence (BLAST_SequenceBlk *seq_blk)
 Adds a specialized representation of sequence data to a sequence block.
Int4 BLAST_TranslateCompressedSequence (Uint1 *translation, Int4 length, const Uint1 *nt_seq, Int2 frame, Uint1 *prot_seq)
 Translate a nucleotide sequence without ambiguity codes.
Int2 GetReverseNuclSequence (const Uint1 *sequence, Int4 length, Uint1 **rev_sequence_ptr)
 Reverse a nucleotide sequence in the blastna encoding, adding sentinel bytes on both ends.
Int1 BLAST_ContextToFrame (EBlastProgramType prog_number, Uint4 context_number)
 This function translates the context number of a context into the frame of the sequence.
Int2 BLAST_PackDNA (const Uint1 *buffer, Int4 length, EBlastEncoding encoding, Uint1 **packed_seq)
 Convert a sequence in ncbi4na or blastna encoding into a packed sequence in ncbi2na encoding.
size_t BLAST_GetTranslatedProteinLength (size_t nucleotide_length, unsigned int context)
 Calculates the length of frame for a translated protein.
Int2 BLAST_CreateMixedFrameDNATranslation (BLAST_SequenceBlk *query_blk, const BlastQueryInfo *query_info)
 Initialize the mixed-frame sequence for out-of-frame gapped extension.
static Uint1 * s_BlastGetTranslationTable (const Uint1 *genetic_code, Boolean reverse_complement)
 Gets the translation array for a given genetic code.
Int2 BLAST_GetAllTranslations (const Uint1 *nucl_seq, EBlastEncoding encoding, Int4 nucl_length, const Uint1 *genetic_code, Uint1 **translation_buffer_ptr, Int4 **frame_offsets_ptr, Uint1 **mixed_seq_ptr)
 Translate nucleotide into 6 frames.
int Blast_GetPartialTranslation (const Uint1 *nucl_seq, Int4 nucl_length, Int2 frame, const Uint1 *genetic_code, Uint1 **translation_buffer_ptr, Int4 *protein_length, Uint1 **mixed_seq_ptr)
 Get one frame translation - needed when only parts of subject sequences are translated.
Int4 BLAST_FrameToContext (Int2 frame, EBlastProgramType program)
 Convert translation frame or strand into a context number suitable for indexing into the BlastQueryInfo::contexts array.
Int4 BSearchInt4 (Int4 n, Int4 *A, Int4 size)
 The following binary search routine assumes that array A is filled.
SBlastTargetTranslation * BlastTargetTranslationFree (SBlastTargetTranslation *target_t)
 Free SBlastTargetTranslation object to be freed [in].
Int2 BlastTargetTranslationNew (BLAST_SequenceBlk *subject_blk, const Uint1 *gen_code_string, EBlastProgramType program_number, Boolean is_ooframe, SBlastTargetTranslation **target)
 Sets up structure for target translation.
double * BLAST_GetStandardAaProbabilities ()
 Get the standard amino acid probabilities.
char * BLAST_StrToUpper (const char *string)
 Returns a copy of the input string with all its characters turned to uppercase.
unsigned int BLAST_GetNumberOfContexts (EBlastProgramType p)
 Get the number of contexts for a given program.
SBlastProgress * SBlastProgressNew (void *user_data)
 Allocates and initializes a new SBlastProgress structure.
SBlastProgress * SBlastProgressFree (SBlastProgress *progress_info)
 Deallocates a SBlastProgress structure.
void SBlastProgressReset (SBlastProgress *progress_info)
 Resets the progress structure to its original state (as if newly allocated) for a fresh start without touching the user_data field.

Variables

static char const rcsid []

Function Documentation

void __sfree (  void **  x  ) 
 

Implemented in blast_util.c.

See also:
sfree

Definition at line 45 of file blast_util.c.

References free().

Int1 BLAST_ContextToFrame (  EBlastProgramType  prog_number,
Uint4  context_number
) 
 

This function translates the context number of a context into the frame of the sequence.

Parameters:
prog_number Integer corresponding to the BLAST program
context_number Context number
Returns:
Sequence frame: -1,1 for nucleotides, -3,-2,-1,1,2,3 for translations, 0 for proteins and INT1_MAX in case of unsupported program

Definition at line 813 of file blast_util.c.

References Blast_QueryIsProtein(), eBlastTypeBlastn, eBlastTypeBlastx, eBlastTypePhiBlastn, eBlastTypeRpsTblastn, eBlastTypeTblastx, INT1_MAX, NUM_FRAMES, and NUM_STRANDS.

Referenced by BLAST_GetAllTranslations(), BlastMaskLocDNAToProtein(), BlastMaskLocProteinToDNA(), BlastQueryInfoNew(), BlastTargetTranslationNew(), BOOST_AUTO_TEST_CASE(), OffsetArrayToContextOffsets(), s_AddMask(), and s_BlastSearchEngineCore().

Int2 BLAST_CreateMixedFrameDNATranslation (  BLAST_SequenceBlk *  query_blk,
const BlastQueryInfo *  query_info
) 
 

Initialize the mixed-frame sequence for out-of-frame gapped extension.

Parameters:
query_blk Sequence block containing the concatenated frames of the query. The mixed-frame sequence is saved here. [in] [out]
query_info Query information structure containing offsets into the* concatenated sequence. [in]

Definition at line 905 of file blast_util.c.

References buffer, CODON_LENGTH, BlastQueryInfo::contexts, BlastQueryInfo::last_context, malloc(), NULLB, BlastContextInfo::query_length, BlastContextInfo::query_offset, and QueryInfo_GetSeqBufLen().

Referenced by BLAST_MainSetUp(), and BOOST_AUTO_TEST_CASE().

Int4 BLAST_FrameToContext (  Int2  frame,
EBlastProgramType  program
) 
 

Convert translation frame or strand into a context number suitable for indexing into the BlastQueryInfo::contexts array.

Parameters:
frame Frame (allowed values: 1,2,3,-1,-2,-3, 0) [in]
program Type of BLAST program [in]
Returns:
context number: 0 or 1 for nucleotide query/subjects, a value between 0 and 5 (inclusive) for translated query/subjects, and 0 for protein query/subjects.

Definition at line 1174 of file blast_util.c.

References ASSERT, Blast_QueryIsNucleotide(), Blast_QueryIsTranslated(), Blast_SubjectIsNucleotide(), and Blast_SubjectIsTranslated().

Referenced by BLAST_GetGappedScore(), Blast_HSPGetTargetTranslation(), BLAST_SmithWatermanGetGappedScore(), BlastRPSWordFinder(), and s_BlastHSPListRPSUpdate().

Int2 BLAST_GetAllTranslations (  const Uint1 *  nucl_seq,
EBlastEncoding  encoding,
Int4  nucl_length,
const Uint1 *  genetic_code,
Uint1 **  translation_buffer_ptr,
Int4 **  frame_offsets_ptr,
Uint1 **  mixed_seq_ptr
) 
 

Translate nucleotide into 6 frames.

All frames are put into a translation buffer, with sentinel NULLB bytes in between. Array of offsets into the translation buffer is also returned. For out-of-frame gapping option, a mixed frame sequence is created.

Parameters:
nucl_seq The nucleotide sequence [in]
encoding Sequence encoding: ncbi2na or ncbi4na [in]
nucl_length Length of the nucleotide sequence [in]
genetic_code The genetic code to be used for translations, in ncbistdaa encoding [in]
translation_buffer_ptr Buffer to hold the frames of the translated sequence. [out]
frame_offsets_ptr Offsets into the translation buffer for each frame. [out]
mixed_seq_ptr Pointer to buffer for the mixed frame sequence [out]

Definition at line 1016 of file blast_util.c.

References BLAST_ContextToFrame(), BLAST_TranslateCompressedSequence(), eBlastEncodingNcbi2na, eBlastEncodingNcbi4na, eBlastTypeBlastx, FALSE, GetReverseNuclSequence(), malloc(), NUM_FRAMES, s_BlastGetTranslationTable(), and TRUE.

Referenced by BlastTargetTranslationNew(), BOOST_AUTO_TEST_CASE(), and s_BlastSearchEngineCore().

unsigned int BLAST_GetNumberOfContexts (  EBlastProgramType  program  ) 
 

Get the number of contexts for a given program.

This corresponds to the number of translation frames or strands whenever applicable.

Returns:
0 on unsupported program, non-zero otherwise

Definition at line 1336 of file blast_util.c.

References Blast_ProgramIsValid(), Blast_QueryIsNucleotide(), Blast_QueryIsTranslated(), NUM_FRAMES, and NUM_STRANDS.

Referenced by BlastHSPStreamMerge(), BlastQueryInfoGetEffSearchSpace(), BlastQueryInfoGetQueryLength(), BlastQueryInfoNew(), BlastQueryInfoSetEffSearchSpace(), BlastSetUp_GetFilteringLocations(), CBlastAncillaryData::CBlastAncillaryData(), GetNumberOfContexts(), and CBlastQueryFilteredFrames::GetNumFrames().

int Blast_GetPartialTranslation (  const Uint1 *  nucl_seq,
Int4  nucl_length,
Int2  frame,
const Uint1 *  genetic_code,
Uint1 **  translation_buffer_ptr,
Int4 *  protein_length,
Uint1 **  mixed_seq_ptr
) 
 

Get one frame translation - needed when only parts of subject sequences are translated.

Parameters:
nucl_seq Pointer to start of nucleotide sequence to be translated [in]
nucl_length Length of nucleotide sequence to be translated [in]
frame What frame to translate into [in]
genetic_code What genetic code to use? [in]
translation_buffer_ptr Pointer to buffer with translated sequence [out]
protein_length Length of the translation buffer [out]
mixed_seq_ptr Pointer to buffer with mixed frame sequence, in case of out-of-frame gapping; buffer filled only if argument not NULL. [out]

Definition at line 1110 of file blast_util.c.

References BLAST_GetTranslation(), CODON_LENGTH, GetReverseNuclSequence(), and malloc().

Referenced by Blast_HSPGetPartialSubjectTranslation(), and s_SequenceGetTranslatedRange().

double* BLAST_GetStandardAaProbabilities (  void   ) 
 

Get the standard amino acid probabilities.

This is basically a wrapper for BlastScoreBlkNew() and Blast_ResFreqStdComp() from blast_stat.c with a more intention-revealing name :) Caller is responsible for deallocating return value via sfree().

Returns:
NULL if there is not enough memory otherwise an array of length BLASTAA_SIZE, where each index corresponds to an amino acid as specified in the NCBIstdaa encoding.

Definition at line 1286 of file blast_util.c.

References Blast_ResFreqFree(), Blast_ResFreqNew(), Blast_ResFreqStdComp(), BLASTAA_SEQ_CODE, BLASTAA_SIZE, malloc(), Blast_ResFreq::prob, and TRUE.

Referenced by _PSISequenceWeightsNew(), ColumnResidueProfile::getBackgroundResFreq(), GetStandardProbability(), Kappa_compactSearchItemsNew(), and PSICreatePssmFromFrequencyRatios().

size_t BLAST_GetTranslatedProteinLength (  size_t  nucleotide_length,
unsigned int  context
) 
 

Calculates the length of frame for a translated protein.

Parameters:
nucleotide_length Length of the nucleotide sequence translated [in]
context Index of the translated frame (values: 0 to 5, inclusive) [in]
Returns:
The requested length, or 0 if the nucleotide length is 0

Definition at line 897 of file blast_util.c.

References CODON_LENGTH.

Int4 BLAST_GetTranslation (  const Uint1 *  query_seq,
const Uint1 *  query_seq_rev,
Int4  nt_length,
Int2  frame,
Uint1 *  buffer,
const Uint1 *  genetic_code
) 
 

GetTranslation to get the translation of the nucl.

sequence in the appropriate frame and with the appropriate GeneticCode. The function return an allocated char*, the caller must delete this. The first and last spaces of this char* contain NULLB's.

Parameters:
query_seq Forward strand of the nucleotide sequence [in]
query_seq_rev Reverse strand of the nucleotide sequence [in]
nt_length Length of the nucleotide sequence [in]
frame What frame to translate into? [in]
buffer Preallocated buffer for the translated sequence [in][out]
genetic_code Genetic code to use for translation, in ncbistdaa encoding [in]
Returns:
Length of the translated protein sequence.

Definition at line 409 of file blast_util.c.

References ABS, CODON_LENGTH, FENCE_SENTRY, IS_residue, NULLB, and s_CodonToAA().

Referenced by Blast_GetPartialTranslation(), Blast_HSPGetTargetTranslation(), and BlastTargetTranslationNew().

Int2 BLAST_PackDNA (  const Uint1 *  buffer,
Int4  length,
EBlastEncoding  encoding,
Uint1 **  packed_seq
) 
 

Convert a sequence in ncbi4na or blastna encoding into a packed sequence in ncbi2na encoding.

Needed for 2 sequences BLASTn comparison.

Parameters:
buffer original sequence data (one base per byte) [in]
length length of the sequence data above [in]
encoding source encoding of the sequence data above [in]
packed_seq output buffer containing compressed sequence. Its length will be (length/COMPRESSION_RATIO + 1), caller is responsible for deallocating it [out]
Returns:
0 in case of success, -1 in case of memory allocation failure

Definition at line 844 of file blast_util.c.

References COMPRESSION_RATIO, eBlastEncodingNucleotide, malloc(), NCBI2NA_MASK, and NCBI4NA_TO_BLASTNA.

char* BLAST_StrToUpper (  const char *  string  ) 
 

Returns a copy of the input string with all its characters turned to uppercase.

Useful for saving score matrix names. Caller is responsible for deallocating return value.

Parameters:
string string to copy [in]
Returns:
newly allocated string in upper case or NULL if string is NULL or out of memory

Definition at line 1315 of file blast_util.c.

References NULLB, and strdup.

Referenced by Blast_ScoreBlkMatrixInit().

Int4 BLAST_TranslateCompressedSequence (  Uint1 *  translation,
Int4  length,
const Uint1 *  nt_seq,
Int2  frame,
Uint1 *  prot_seq
) 
 

Translate a nucleotide sequence without ambiguity codes.

This is used for the first-pass translation of the database. The genetic code to be used is determined by the translation_table This function translates a packed (ncbi2na) nucl. alphabet. It views a basepair as being in one of four sets of 2-bits: |0|1|2|3||0|1|2|3||0|1|2|3||...

1st byte | 2 byte | 3rd byte...

A codon that starts at the beginning of the above sequence starts in state "0" and includes basepairs 0, 1, and 2. The next codon, in the same frame, after that starts in state "3" and includes 3, 0, and 1.

Optimization: changed the single main loop to

  • advance to state 0,
  • optimized inner loop does two (3 byte->4 codon) translation per iteration (loads are moved earlier so they can be done in advance.)
  • do remainder

Parameters:
translation The translation table [in]
length Length of the nucleotide sequence [in]
nt_seq The original nucleotide sequence [in]
frame What frame to translate to? [in]
prot_seq Preallocated buffer for the (translated) protein sequence, with NULLB sentinels on either end. [out]

Definition at line 489 of file blast_util.c.

References ABS, CODON_LENGTH, and NULLB.

Referenced by BLAST_GetAllTranslations().

Int2 BlastCompressBlastnaSequence (  BLAST_SequenceBlk *  seq_blk  ) 
 

Adds a specialized representation of sequence data to a sequence block.

In the specialized representation, the byte at offset i packs together nucleotide bases i to i+3

Parameters:
seq_blk structure containing sequence data. Data is assumed to be in blastna format [in][out]

Definition at line 440 of file blast_util.c.

References BLAST_SequenceBlk::compressed_nuc_seq, BLAST_SequenceBlk::compressed_nuc_seq_start, len, BLAST_SequenceBlk::length, malloc(), MIN, and BLAST_SequenceBlk::sequence.

Int2 BlastNumber2Program (  EBlastProgramType  number,
char **  program
) 
 

Return string name for program given a number.

Return is zero on success.

Parameters:
number Enumerated value of program [in]
program string name of program (memory should be deallocated by called) [out]

Definition at line 296 of file blast_util.c.

References eBlastTypeBlastn, eBlastTypeBlastp, eBlastTypeBlastx, eBlastTypePhiBlastn, eBlastTypePhiBlastp, eBlastTypePsiBlast, eBlastTypePsiTblastn, eBlastTypeRpsBlast, eBlastTypeRpsTblastn, eBlastTypeTblastn, eBlastTypeTblastx, and strdup.

Referenced by Blast_ProgramNameFromType(), and CRedoAlignmentTestFixture::runRedoAlignmentCoreUnitTest().

Int2 BlastProgram2Number (  const char *  program,
EBlastProgramType *  number
) 
 

Set number for a given program type.

Return is zero on success.

Parameters:
program string name of program [in]
number Enumerated value of program [out]

Definition at line 264 of file blast_util.c.

References eBlastTypeBlastn, eBlastTypeBlastp, eBlastTypeBlastx, eBlastTypePhiBlastn, eBlastTypePhiBlastp, eBlastTypePsiBlast, eBlastTypePsiTblastn, eBlastTypeRpsBlast, eBlastTypeRpsTblastn, eBlastTypeTblastn, eBlastTypeTblastx, eBlastTypeUndefined, and strcasecmp.

Referenced by NetworkProgram2BlastProgramType().

Int2 BlastSeqBlkNew (  BLAST_SequenceBlk **  retval  ) 
 

Allocates a new sequence block structure.

Parameters:
retval Pointer to where the sequence block structure will be allocated [out]

Definition at line 132 of file blast_util.c.

References calloc().

Referenced by AascanTestFixture::AascanTestFixture(), BlastSetUp_SeqBlkNew(), BOOST_AUTO_TEST_CASE(), AalookupTestFixture::GetSeqBlk(), InitializeBlastScoreBlk(), s_SetupSequencesForGappedReevaluateTest(), s_SetupSequencesForUngappedReevaluateNucl(), s_SetupSequencesForUngappedReevaluateTransl(), TestFixture::SetUpQuery(), TestFixture::SetUpSubject(), SetupSubjects_OMF(), CPssmEngine::x_InitializeScoreBlock(), and CPhiblastTestFixture::x_SetupSequenceBlk().

Int2 BlastSeqBlkSetCompressedSequence (  BLAST_SequenceBlk *  seq_blk,
const Uint1 *  sequence
) 
 

Stores the compressed nucleotide sequence in the sequence block structure for the subject sequence when BLASTing 2 sequences.

This sequence should be encoded in eBlastEncodingNcbi2na and NOT have sentinel bytes (as this encoding doesn't allow them).

Parameters:
seq_blk The sequence block structure to modify [in/out]
sequence Actual sequence buffer. [in]

Definition at line 163 of file blast_util.c.

References BLAST_SequenceBlk::oof_sequence, BLAST_SequenceBlk::sequence, BLAST_SequenceBlk::sequence_allocated, and TRUE.

Referenced by TestFixture::SetUpSubject().

Int2 BlastSeqBlkSetSeqRanges (  BLAST_SequenceBlk *  seq_blk,
SSeqRange *  seq_ranges,
Uint4  num_seq_ranges,
Boolean  copy_seq_ranges
) 
 

Sets the seq_range and related fields appropriately in the BLAST_SequenceBlk structure.

Parameters:
seq_blk The sequence block structure to modify [in/out]
seq_ranges sequence ranges to copy [in]
num_seq_ranges number of elements in array above [in]
copy_seq_ranges set to TRUE if seq_ranges should be copied to the BLAST_SequenceBlk and assume its ownership, set to FALSE if the pointer should be copied and the ownership of the seq_ranges remains in the caller's possession.
Note:
this function will free the memory previously allocated to BLAST_SequenceBlk::seq_ranges (if applicable) and overwrite it with the seq_ranges argument. This might invalidate BLAST_SequenceBlk structures that were copied off of this one.

Definition at line 178 of file blast_util.c.

References ASSERT, calloc(), FALSE, BLAST_SequenceBlk::num_seq_ranges, s_BlastSequenceBlkFreeSeqRanges(), BLAST_SequenceBlk::seq_ranges, BLAST_SequenceBlk::seq_ranges_allocated, and TRUE.

Referenced by BOOST_AUTO_TEST_CASE(), SetupSubjects_OMF(), and TestFixture::SkipMaskedRangesCore().

Int2 BlastSeqBlkSetSequence (  BLAST_SequenceBlk *  seq_blk,
const Uint1 *  sequence,
Int4  seqlen
) 
 

Stores the sequence in the sequence block structure.

Parameters:
seq_blk The sequence block structure to modify [in/out]
sequence Actual sequence buffer. The first byte must be a sentinel byte [in]
seqlen Length of the sequence buffer above [in]

Definition at line 146 of file blast_util.c.

References BLAST_SequenceBlk::length, BLAST_SequenceBlk::oof_sequence, BLAST_SequenceBlk::sequence, BLAST_SequenceBlk::sequence_start, BLAST_SequenceBlk::sequence_start_allocated, and TRUE.

Referenced by AascanTestFixture::AascanTestFixture(), BOOST_AUTO_TEST_CASE(), AalookupTestFixture::GetSeqBlk(), InitializeBlastScoreBlk(), s_SetupSequencesForGappedReevaluateTest(), s_SetupSequencesForUngappedReevaluateNucl(), s_SetupSequencesForUngappedReevaluateTransl(), TestFixture::SetUpQuery(), CPssmEngine::x_InitializeScoreBlock(), and CPhiblastTestFixture::x_SetupSequenceBlk().

void BlastSequenceBlkClean (  BLAST_SequenceBlk *  seq_blk  ) 
 

Deallocate memory only for the sequence in the sequence block.

Definition at line 210 of file blast_util.c.

References FALSE, BLAST_SequenceBlk::oof_sequence, BLAST_SequenceBlk::oof_sequence_allocated, s_BlastSequenceBlkFreeSeqRanges(), BLAST_SequenceBlk::sequence, BLAST_SequenceBlk::sequence_allocated, BLAST_SequenceBlk::sequence_start, BLAST_SequenceBlk::sequence_start_allocated, and sfree.

Referenced by BLAST_ComputeTraceback(), and BlastSequenceBlkFree().

void BlastSequenceBlkCopy (  BLAST_SequenceBlk **  copy,
BLAST_SequenceBlk *  src
) 
 

Copies contents of the source sequence block without copying sequence buffers; sets all "field_allocated" booleans to FALSE, to make sure fields are not freed on the call to BlastSequenceBlkFree.

Parameters:
copy New sequence block [out]
src Input sequence block [in]

Definition at line 245 of file blast_util.c.

References ASSERT, BlastMemDup(), and FALSE.

Referenced by s_MultiSeqGetSequence(), and s_QueryFactoryGetSequence().

BLAST_SequenceBlk* BlastSequenceBlkFree (  BLAST_SequenceBlk *  seq_blk  ) 
 

Deallocate memory for a sequence block.

Definition at line 231 of file blast_util.c.

References BlastMaskLocFree(), BlastSequenceBlkClean(), BLAST_SequenceBlk::compressed_nuc_seq_start, BLAST_SequenceBlk::lcase_mask, BLAST_SequenceBlk::lcase_mask_allocated, and sfree.

Referenced by BOOST_AUTO_TEST_CASE(), s_MatchingSequenceRelease(), s_RPSComputeTraceback(), TestFixture::TearDownQuery(), TestFixture::TearDownSubject(), AalookupTestFixture::~AalookupTestFixture(), AascanTestFixture::~AascanTestFixture(), CMultiSeqInfo::~CMultiSeqInfo(), and CQueryFactoryInfo::~CQueryFactoryInfo().

Int2 BlastSetUp_SeqBlkNew (  const Uint1 *  buffer,
Int4  length,
BLAST_SequenceBlk **  seq_blk,
Boolean  buffer_allocated
) 
 

Allocates memory for *sequence_blk and then populates it.

Parameters:
buffer start of sequence [in]
length query sequence length [in]
seq_blk SequenceBlk to be allocated and filled in [out]
buffer_allocated Is the buffer allocated? If yes, 'sequence_start' is the start of the sequence, otherwise it is 'sequence'. [in]

Definition at line 106 of file blast_util.c.

References ASSERT, BlastSeqBlkNew(), and TRUE.

Referenced by AalookupTestFixture::GetSeqBlk().

SBlastTargetTranslation* BlastTargetTranslationFree (  SBlastTargetTranslation *  target_t  ) 
 

Free SBlastTargetTranslation object to be freed [in].

Definition at line 1211 of file blast_util.c.

References SBlastTargetTranslation::num_frames, SBlastTargetTranslation::range, sfree, and SBlastTargetTranslation::translations.

Int2 BlastTargetTranslationNew (  BLAST_SequenceBlk *  subject_blk,
const Uint1 *  gen_code_string,
EBlastProgramType  program_number,
Boolean  is_ooframe,
SBlastTargetTranslation **  target
) 
 

Sets up structure for target translation.

Target sequence information [in] Genetic code translation information [in] BLAST program [in] Out-of-frame translation if true [in] Structure being set up. [out]

Definition at line 1231 of file blast_util.c.

References BLAST_ContextToFrame(), BLAST_GetAllTranslations(), BLAST_GetTranslation(), calloc(), eBlastEncodingNcbi4na, eBlastTypeBlastx, SBlastTargetTranslation::gen_code_string, GetReverseNuclSequence(), BLAST_SequenceBlk::length, malloc(), NUM_FRAMES, SBlastTargetTranslation::num_frames, BLAST_SequenceBlk::oof_sequence, BLAST_SequenceBlk::oof_sequence_allocated, SBlastTargetTranslation::partial, SBlastTargetTranslation::program_number, BLAST_SequenceBlk::sequence_start, sfree, SBlastTargetTranslation::translations, and TRUE.

Referenced by Blast_HSPListReevaluateWithAmbiguitiesUngapped(), Blast_TracebackFromHSPList(), BOOST_AUTO_TEST_CASE(), and s_ComputeNumIdentities().

Int4 BSearchInt4 (  Int4  n,
Int4 *  A,
Int4  size
) 
 

The following binary search routine assumes that array A is filled.

Definition at line 1194 of file blast_util.c.

Int2 GetReverseNuclSequence (  const Uint1 *  sequence,
Int4  length,
Uint1 **  rev_sequence_ptr
) 
 

Reverse a nucleotide sequence in the blastna encoding, adding sentinel bytes on both ends.

Parameters:
sequence Forward strand of the sequence [in]
length Length of the sequence plus 1 for the sentinel byte [in]
rev_sequence_ptr Reverse strand of the sequence [out]

Definition at line 788 of file blast_util.c.

References malloc(), and NULLB.

Referenced by BLAST_GetAllTranslations(), Blast_GetPartialTranslation(), Blast_HSPGetTargetTranslation(), and BlastTargetTranslationNew().

static Uint1* s_BlastGetTranslationTable (  const Uint1 *  genetic_code,
Boolean  reverse_complement
)  [static]
 

Gets the translation array for a given genetic code.

This array is optimized for the NCBI2na alphabet. The reverse complement can also be spcified.

Parameters:
genetic_code Genetic code string in ncbistdaa encoding [in]
reverse_complement Get translation table for the reverse strand? [in]
Returns:
The translation table.

Definition at line 961 of file blast_util.c.

References calloc().

Referenced by BLAST_GetAllTranslations().

static void s_BlastSequenceBlkFreeSeqRanges (  BLAST_SequenceBlk *  seq_blk  )  [static]
 

Auxiliary function to free the BLAST_SequenceBlk::seq_ranges field if applicable.

Parameters:
seq_blk The sequence block structure to manipulate [in|out]

Definition at line 95 of file blast_util.c.

References ASSERT, FALSE, BLAST_SequenceBlk::num_seq_ranges, BLAST_SequenceBlk::seq_ranges, BLAST_SequenceBlk::seq_ranges_allocated, and sfree.

Referenced by BlastSeqBlkSetSeqRanges(), and BlastSequenceBlkClean().

static Uint1 s_CodonToAA (  Uint1 *  codon,
const Uint1 *  codes
)  [static]
 

Translate 3 nucleotides into an amino acid MUST have 'X' as unknown amino acid.

Parameters:
codon 3 values in ncbi4na code
codes Geneic code string to use (must be in ncbistdaa encoding!)
Returns:
Amino acid in ncbistdaa

Definition at line 351 of file blast_util.c.

References AMINOACID_TO_NCBISTDAA, and FENCE_SENTRY.

Referenced by BLAST_GetTranslation().

SBlastProgress* SBlastProgressFree (  SBlastProgress *  progress_info  ) 
 

Deallocates a SBlastProgress structure.

Implemented in blast_util.c

Definition at line 1360 of file blast_util.c.

References sfree.

SBlastProgress* SBlastProgressNew (  void *  user_data  ) 
 

Allocates and initializes a new SBlastProgress structure.

Parameters:
user_data user-provided data (not owned by the resulting structure) [in] Implemented in blast_util.c

Definition at line 1350 of file blast_util.c.

References calloc(), and SBlastProgress::user_data.

Referenced by CBl2Seq::SetInterruptCallback().

void SBlastProgressReset (  SBlastProgress *  progress_info  ) 
 

Resets the progress structure to its original state (as if newly allocated) for a fresh start without touching the user_data field.

Definition at line 1369 of file blast_util.c.

References ePrelimSearch, and SBlastProgress::stage.

Referenced by CBl2Seq::RunFullSearch().

Int4 SSeqRangeArrayLessThanOrEqual (  const SSeqRange *  ranges,
Int4  num_ranges,
Int4  target
) 
 

Returns the index of the range, such that this element is the first range that either contains the target or if no such range exists, the index of the first range, such that the target is less than this range.

Precondition:
ranges array is sorted on the starting coordinates (i.e.: SSeqRange::left)
Parameters:
ranges array of SSeqRange structures to search [in]
num_ranges number of elements in the ranges array [in]
target element to look for [in]
Returns:
the index of interest in the ranges array or -1 if the function was called with invalid parameters

Definition at line 61 of file blast_util.c.

Referenced by BOOST_AUTO_TEST_CASE(), and s_AdjustSubjectChunks().

SSeqRange SSeqRangeNew (  Int4  start,
Int4  stop
) 
 

Create a new SSeqRange structure with both fields initialized.

Parameters:
start the start of the range [in]
stop the end of the range [in]

Definition at line 52 of file blast_util.c.

References SSeqRange::left, and SSeqRange::right.

Referenced by BOOST_AUTO_TEST_CASE().

Variable Documentation

char const rcsid[] [static]
 

Initial value:

 
    "$Id: blast_util.c 177349 2009-11-30 19:00:28Z coulouri $"

Definition at line 36 of file blast_util.c.

Generated on Wed Dec 9 07:41:17 2009 for NCBI C++ ToolKit by  doxygen 1.4.6
Modified on Wed Dec 09 08:18:30 2009 by modify_doxy.py rev. 173732