libStatGen Software
1
|
This class represents the CIGAR without any methods to set the cigar (see CigarRoller for that). More...
#include <Cigar.h>
Classes | |
struct | CigarOperator |
Public Types | |
enum | Operation { none = 0, match, mismatch, insert, del, skip, softClip, hardClip, pad } |
Enum for the cigar operations. More... | |
Public Member Functions | |
Cigar () | |
Default constructor initializes as a CIGAR with no operations. | |
void | getCigarString (String &cigarString) const |
Set the passed in String to the string reprentation of the Cigar operations in this object. | |
void | getCigarString (std::string &cigarString) const |
Set the passed in std::string to the string reprentation of the Cigar operations in this object. | |
void | getExpandedString (std::string &s) const |
Sets the specified string to a valid CIGAR string of characters that represent the cigar with no digits (a CIGAR of "3M" would return "MMM"). | |
const CigarOperator & | operator[] (int i) const |
Return the Cigar Operation at the specified index (starting at 0). | |
const CigarOperator & | getOperator (int i) const |
Return the Cigar Operation at the specified index (starting at 0). | |
bool | operator== (Cigar &rhs) const |
Return true if the 2 Cigars are the same (the same operations of the same sizes). | |
int | size () const |
Return the number of cigar operations. | |
void | Dump () const |
Write this object as a string to cout. | |
int | getExpectedQueryBaseCount () const |
Return the length of the read that corresponds to the current CIGAR string. | |
int | getExpectedReferenceBaseCount () const |
Return the number of bases in the reference that this CIGAR "spans". | |
int | getNumBeginClips () const |
Return the number of clips that are at the beginning of the cigar. | |
int | getNumEndClips () const |
Return the number of clips that are at the end of the cigar. | |
int32_t | getRefOffset (int32_t queryIndex) |
Return the reference offset associated with the specified query index or INDEX_NA based on this cigar. | |
int32_t | getQueryIndex (int32_t refOffset) |
Return the query index associated with the specified reference offset or INDEX_NA based on this cigar. | |
int32_t | getRefPosition (int32_t queryIndex, int32_t queryStartPos) |
Return the reference position associated with the specified query index or INDEX_NA based on this cigar and the specified queryStartPos which is the leftmost mapping position of the first matching base in the query. | |
int32_t | getQueryIndex (int32_t refPosition, int32_t queryStartPos) |
Return the query index or INDEX_NA associated with the specified reference offset when the query starts at the specified reference position. | |
int32_t | getExpandedCigarIndexFromQueryIndex (int32_t queryIndex) |
Returns the index into the expanded cigar for the cigar associated with the specified queryIndex. | |
int32_t | getExpandedCigarIndexFromRefOffset (int32_t refOffset) |
Returns the index into the expanded cigar for the cigar associated with the specified reference offset. | |
int32_t | getExpandedCigarIndexFromRefPos (int32_t refPosition, int32_t queryStartPos) |
Returns the index into the expanded cigar for the cigar associated with the specified reference position and queryStartPos. | |
char | getCigarCharOp (int32_t expandedCigarIndex) |
Return the character code of the cigar operator associated with the specified expanded CIGAR index. | |
char | getCigarCharOpFromQueryIndex (int32_t queryIndex) |
Return the character code of the cigar operator associated with the specified queryIndex. | |
char | getCigarCharOpFromRefOffset (int32_t refOffset) |
Return the character code of the cigar operator associated with the specified reference offset. | |
char | getCigarCharOpFromRefPos (int32_t refPosition, int32_t queryStartPos) |
Return the character code of the cigar operator associated with the specified reference position. | |
uint32_t | getNumOverlaps (int32_t start, int32_t end, int32_t queryStartPos) |
Return the number of bases that overlap the reference and the read associated with this cigar that falls within the specified region. | |
bool | hasIndel () |
Return whether or not the cigar has indels (insertions or delections) | |
Static Public Member Functions | |
static bool | foundInReference (Operation op) |
Return true if the specified operation is found in the reference sequence, false if not. | |
static bool | foundInReference (char op) |
Return true if the specified operation is found in the reference sequence, false if not. | |
static bool | foundInReference (const CigarOperator &op) |
Return true if the specified operation is found in the reference sequence, false if not. | |
static bool | foundInQuery (Operation op) |
Return true if the specified operation is found in the query sequence, false if not. | |
static bool | foundInQuery (char op) |
Return true if the specified operation is found in the query sequence, false if not. | |
static bool | foundInQuery (const CigarOperator &op) |
Return true if the specified operation is found in the query sequence, false if not. | |
static bool | isClip (Operation op) |
Return true if the specified operation is a clipping operation, false if not. | |
static bool | isClip (char op) |
Return true if the specified operation is a clipping operation, false if not. | |
static bool | isClip (const CigarOperator &op) |
Return true if the specified operation is a clipping operation, false if not. | |
static bool | isMatchOrMismatch (Operation op) |
Return true if the specified operation is a match/mismatch operation, false if not. | |
static bool | isMatchOrMismatch (const CigarOperator &op) |
Return true if the specified operation is a match/mismatch operation, false if not. | |
Static Public Attributes | |
static const int | MAX_OP_VALUE = pad |
static const int32_t | INDEX_NA = -1 |
Value associated with an index that is not applicable/does not exist, used for converting between query and reference indexes/offsets when an associated index/offset does not exist. | |
Protected Member Functions | |
void | clearQueryAndReferenceIndexes () |
void | setQueryAndReferenceIndexes () |
Protected Attributes | |
std::vector< CigarOperator > | cigarOperations |
Friends | |
std::ostream & | operator<< (std::ostream &stream, const Cigar &cigar) |
Writes all of the cigar operations contained in the cigar to the passed in stream. |
This class represents the CIGAR without any methods to set the cigar (see CigarRoller for that).
This class represents the CIGAR. It contains methods for converting to strings and extracting information from the cigar on how a read maps to the reference.
It only contains read only methods. There are no ways to set values. To set a value, a child class must be used.
enum Cigar::Operation |
Enum for the cigar operations.
Definition at line 87 of file Cigar.h.
{ none=0, ///< no operation has been set. match, ///< match/mismatch operation. Associated with CIGAR Operation "M" mismatch, ///< mismatch operation. Associated with CIGAR Operation "M" insert, ///< insertion to the reference (the query sequence contains bases that have no corresponding base in the reference). Associated with CIGAR Operation "I" del, ///< deletion from the reference (the reference contains bases that have no corresponding base in the query sequence). Associated with CIGAR Operation "D" skip, ///< skipped region from the reference (the reference contains bases that have no corresponding base in the query sequence). Associated with CIGAR Operation "N" softClip, ///< Soft clip on the read (clipped sequence present in the query sequence, but not in reference). Associated with CIGAR Operation "S" hardClip, ///< Hard clip on the read (clipped sequence not present in the query sequence or reference). Associated with CIGAR Operation "H" pad ///< Padding (not in reference or query). Associated with CIGAR Operation "P" };
static bool Cigar::foundInQuery | ( | Operation | op | ) | [inline, static] |
Return true if the specified operation is found in the query sequence, false if not.
Definition at line 219 of file Cigar.h.
References insert, match, mismatch, and softClip.
Referenced by foundInQuery(), SamRecord::shiftIndelsLeft(), SamFilter::softClip(), CigarHelper::softClipBeginByRefPos(), and CigarHelper::softClipEndByRefPos().
static bool Cigar::foundInQuery | ( | char | op | ) | [inline, static] |
static bool Cigar::foundInQuery | ( | const CigarOperator & | op | ) | [inline, static] |
Return true if the specified operation is found in the query sequence, false if not.
Definition at line 254 of file Cigar.h.
References foundInQuery().
{ return(foundInQuery(op.operation)); }
static bool Cigar::foundInReference | ( | Operation | op | ) | [inline, static] |
Return true if the specified operation is found in the reference sequence, false if not.
Definition at line 177 of file Cigar.h.
References del, match, mismatch, and skip.
Referenced by foundInReference(), CigarHelper::softClipBeginByRefPos(), and CigarHelper::softClipEndByRefPos().
static bool Cigar::foundInReference | ( | char | op | ) | [inline, static] |
static bool Cigar::foundInReference | ( | const CigarOperator & | op | ) | [inline, static] |
Return true if the specified operation is found in the reference sequence, false if not.
Definition at line 212 of file Cigar.h.
References foundInReference().
{ return(foundInReference(op.operation)); }
char Cigar::getCigarCharOp | ( | int32_t | expandedCigarIndex | ) |
Return the character code of the cigar operator associated with the specified expanded CIGAR index.
'?' is returned for an out of range index.
Definition at line 295 of file Cigar.cpp.
Referenced by getCigarCharOpFromQueryIndex(), getCigarCharOpFromRefOffset(), and getCigarCharOpFromRefPos().
{ // Check if the expanded cigar has been set yet if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { // Set the expanded cigar. setQueryAndReferenceIndexes(); } // Check to see if the index is in range. if((expandedCigarIndex < 0) || ((uint32_t)expandedCigarIndex >= myExpandedCigar.length())) { return('?'); } return(myExpandedCigar[expandedCigarIndex]); }
char Cigar::getCigarCharOpFromQueryIndex | ( | int32_t | queryIndex | ) |
Return the character code of the cigar operator associated with the specified queryIndex.
'?' is returned for an out of range index.
Definition at line 314 of file Cigar.cpp.
References getCigarCharOp(), and getExpandedCigarIndexFromQueryIndex().
{ return(getCigarCharOp(getExpandedCigarIndexFromQueryIndex(queryIndex))); }
char Cigar::getCigarCharOpFromRefOffset | ( | int32_t | refOffset | ) |
Return the character code of the cigar operator associated with the specified reference offset.
'?' is returned for an out of range offset.
Definition at line 320 of file Cigar.cpp.
References getCigarCharOp(), and getExpandedCigarIndexFromRefOffset().
{ return(getCigarCharOp(getExpandedCigarIndexFromRefOffset(refOffset))); }
char Cigar::getCigarCharOpFromRefPos | ( | int32_t | refPosition, |
int32_t | queryStartPos | ||
) |
Return the character code of the cigar operator associated with the specified reference position.
'?' is returned for an out of range reference position.
Definition at line 326 of file Cigar.cpp.
References getCigarCharOp(), and getExpandedCigarIndexFromRefPos().
{ return(getCigarCharOp(getExpandedCigarIndexFromRefPos(refPosition, queryStartPos))); }
void Cigar::getCigarString | ( | String & | cigarString | ) | const |
Set the passed in String to the string reprentation of the Cigar operations in this object.
Definition at line 52 of file Cigar.cpp.
Referenced by Dump(), and SamRecord::setCigar().
{ std::string cigar; getCigarString(cigar); cigarString = cigar.c_str(); return; }
void Cigar::getCigarString | ( | std::string & | cigarString | ) | const |
Set the passed in std::string to the string reprentation of the Cigar operations in this object.
Definition at line 36 of file Cigar.cpp.
{ using namespace STLUtilities; std::vector<CigarOperator>::const_iterator i; cigarString.clear(); // clear result string // Progressively append the character representations of the operations to // the cigar string. for (i = cigarOperations.begin(); i != cigarOperations.end(); i++) { cigarString << (*i).count << (*i).getChar(); } }
int32_t Cigar::getExpandedCigarIndexFromQueryIndex | ( | int32_t | queryIndex | ) |
Returns the index into the expanded cigar for the cigar associated with the specified queryIndex.
INDEX_NA returned if the index is out of range.
Definition at line 258 of file Cigar.cpp.
References INDEX_NA.
Referenced by getCigarCharOpFromQueryIndex().
{ // If the vectors aren't set, set them. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } if ((queryIndex < 0) || ((uint32_t)queryIndex >= queryToCigar.size())) { return(INDEX_NA); } return(queryToCigar[queryIndex]); }
int32_t Cigar::getExpandedCigarIndexFromRefOffset | ( | int32_t | refOffset | ) |
Returns the index into the expanded cigar for the cigar associated with the specified reference offset.
INDEX_NA returned if the offset is out of range.
Definition at line 273 of file Cigar.cpp.
References INDEX_NA.
Referenced by getCigarCharOpFromRefOffset(), and getExpandedCigarIndexFromRefPos().
{ // If the vectors aren't set, set them. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } if ((refOffset < 0) || ((uint32_t)refOffset >= refToCigar.size())) { return(INDEX_NA); } return(refToCigar[refOffset]); }
int32_t Cigar::getExpandedCigarIndexFromRefPos | ( | int32_t | refPosition, |
int32_t | queryStartPos | ||
) |
Returns the index into the expanded cigar for the cigar associated with the specified reference position and queryStartPos.
INDEX_NA returned if the position is out of range.
Definition at line 288 of file Cigar.cpp.
References getExpandedCigarIndexFromRefOffset().
Referenced by getCigarCharOpFromRefPos().
{ return(getExpandedCigarIndexFromRefOffset(refPosition - queryStartPos)); }
void Cigar::getExpandedString | ( | std::string & | s | ) | const |
Sets the specified string to a valid CIGAR string of characters that represent the cigar with no digits (a CIGAR of "3M" would return "MMM").
The returned string is actually also a valid CIGAR string. In theory this makes it easier to parse some reads.
Definition at line 63 of file Cigar.cpp.
{ s = ""; std::vector<CigarOperator>::const_iterator i; // Progressively append the character representations of the operations to // the string passed in for (i = cigarOperations.begin(); i != cigarOperations.end(); i++) { for (uint32_t j = 0; j<(*i).count; j++) s += (*i).getChar(); } return; }
int Cigar::getExpectedQueryBaseCount | ( | ) | const |
Return the length of the read that corresponds to the current CIGAR string.
For validation, we should expect that a sequence read in a SAM file will be the same length as the value returned by this method.
Example: 3M2D3M describes a read with three bases matching the reference, then skips 2 bases, then has three more bases that match the reference (match/mismatch). In this case, the read length is expected to be 6.
Example: 3M2I3M describes a read with 3 match/mismatch bases, two extra bases, and then 3 more match/mistmatch bases. The total in this example is 8 bases.
Definition at line 95 of file Cigar.cpp.
References insert, match, mismatch, and softClip.
Referenced by SamValidator::isValidCigar(), and SamFilter::softClip().
{ int matchCount = 0; std::vector<CigarOperator>::const_iterator i; for (i = cigarOperations.begin(); i != cigarOperations.end(); i++) { switch (i->operation) { case match: case mismatch: case softClip: case insert: matchCount += i->count; break; default: // we only care about operations that are in the query sequence. break; } } return matchCount; }
int Cigar::getExpectedReferenceBaseCount | ( | ) | const |
Return the number of bases in the reference that this CIGAR "spans".
When doing range checking, we occassionally need to know how many total bases the CIGAR string represents as compared to the reference.
Examples: 3M2D3M describes a read that overlays 8 bases in the reference. 3M2I3M describes a read with 3 bases that match the reference, two additional bases that aren't in the reference, and 3 more bases that match the reference, so it spans 6 bases in the reference.
Definition at line 120 of file Cigar.cpp.
References del, match, mismatch, and skip.
Referenced by SamTags::createMDTag().
{ int matchCount = 0; std::vector<CigarOperator>::const_iterator i; for (i = cigarOperations.begin(); i != cigarOperations.end(); i++) { switch (i->operation) { case match: case mismatch: case del: case skip: matchCount += i->count; break; default: // we only care about operations that are in the reference sequence. break; } } return matchCount; }
uint32_t Cigar::getNumOverlaps | ( | int32_t | start, |
int32_t | end, | ||
int32_t | queryStartPos | ||
) |
Return the number of bases that overlap the reference and the read associated with this cigar that falls within the specified region.
start | : inclusive 0-based start position (reference position) of the region to check for overlaps in (-1 indicates to start at the beginning of the reference.) |
end | : exclusive 0-based end position (reference position) of the region to check for overlaps in (-1 indicates to go to the end of the reference.) |
queryStartPos | : 0-based leftmost mapping position of the first matcihng base in the query. |
Definition at line 334 of file Cigar.cpp.
References getRefOffset().
Referenced by SamRecord::getNumOverlaps().
{ // Get the overlap info. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } // Get the start and end offsets. int32_t startRefOffset = 0; // If the specified start is more than the queryStartPos, set // the startRefOffset to the appropriate non-zero value. // (if start is <= queryStartPos, than startRefOffset is 0 - it should // not be set to a negative value.) if (start > queryStartPos) { startRefOffset = start - queryStartPos; } int32_t endRefOffset = end - queryStartPos; if (end == -1) { // -1 means that the region goes to the end of the refrerence. // So set endRefOffset to the max refOffset + 1 which is the // size of the refToQuery vector. endRefOffset = refToQuery.size(); } // if endRefOffset is less than 0, then this read does not fall within // the specified region, so return 0. if (endRefOffset < 0) { return(0); } // Get the overlaps for these offsets. // Loop through the read counting positions that match the reference // within this region. int32_t refOffset = 0; int32_t numOverlaps = 0; for (unsigned int queryIndex = 0; queryIndex < queryToRef.size(); queryIndex++) { refOffset = getRefOffset(queryIndex); if (refOffset > endRefOffset) { // Past the end of the specified region, so stop checking // for overlaps since there will be no more. break; } else if ((refOffset >= startRefOffset) && (refOffset < endRefOffset)) { // within the region, increment the counter. ++numOverlaps; } } return(numOverlaps); }
int32_t Cigar::getQueryIndex | ( | int32_t | refOffset | ) |
Return the query index associated with the specified reference offset or INDEX_NA based on this cigar.
Definition at line 202 of file Cigar.cpp.
References INDEX_NA.
Referenced by PileupElementBaseQual::addEntry(), SamRecordHelper::checkSequence(), and SamTags::createMDTag().
{ // If the vectors aren't set, set them. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } if ((refOffset < 0) || ((uint32_t)refOffset >= refToQuery.size())) { return(INDEX_NA); } return(refToQuery[refOffset]); }
int32_t Cigar::getQueryIndex | ( | int32_t | refPosition, |
int32_t | queryStartPos | ||
) |
Return the query index or INDEX_NA associated with the specified reference offset when the query starts at the specified reference position.
Definition at line 240 of file Cigar.cpp.
References INDEX_NA.
{ // If the vectors aren't set, set them. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } int32_t refOffset = refPosition - queryStartPos; if ((refOffset < 0) || ((uint32_t)refOffset >= refToQuery.size())) { return(INDEX_NA); } return(refToQuery[refOffset]); }
int32_t Cigar::getRefOffset | ( | int32_t | queryIndex | ) |
Return the reference offset associated with the specified query index or INDEX_NA based on this cigar.
Definition at line 187 of file Cigar.cpp.
References INDEX_NA.
Referenced by SamQuerySeqWithRefIter::getNextMatchMismatch(), getNumOverlaps(), SamQuerySeqWithRef::seqWithEquals(), and SamQuerySeqWithRef::seqWithoutEquals().
{ // If the vectors aren't set, set them. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } if ((queryIndex < 0) || ((uint32_t)queryIndex >= queryToRef.size())) { return(INDEX_NA); } return(queryToRef[queryIndex]); }
int32_t Cigar::getRefPosition | ( | int32_t | queryIndex, |
int32_t | queryStartPos | ||
) |
Return the reference position associated with the specified query index or INDEX_NA based on this cigar and the specified queryStartPos which is the leftmost mapping position of the first matching base in the query.
Definition at line 217 of file Cigar.cpp.
References INDEX_NA.
Referenced by SamFilter::softClip().
{ // If the vectors aren't set, set them. if ((queryToRef.size() == 0) || (refToQuery.size() == 0)) { setQueryAndReferenceIndexes(); } if ((queryIndex < 0) || ((uint32_t)queryIndex >= queryToRef.size())) { return(INDEX_NA); } if (queryToRef[queryIndex] != INDEX_NA) { return(queryToRef[queryIndex] + queryStartPos); } return(INDEX_NA); }
bool Cigar::hasIndel | ( | ) |
static bool Cigar::isClip | ( | Operation | op | ) | [inline, static] |
Return true if the specified operation is a clipping operation, false if not.
Definition at line 261 of file Cigar.h.
References hardClip, and softClip.
Referenced by isClip(), SamFilter::softClip(), and CigarHelper::softClipEndByRefPos().
static bool Cigar::isClip | ( | char | op | ) | [inline, static] |
static bool Cigar::isClip | ( | const CigarOperator & | op | ) | [inline, static] |
static bool Cigar::isMatchOrMismatch | ( | Operation | op | ) | [inline, static] |
Return true if the specified operation is a match/mismatch operation, false if not.
Definition at line 298 of file Cigar.h.
References match, and mismatch.
Referenced by isMatchOrMismatch(), and SamRecord::shiftIndelsLeft().
static bool Cigar::isMatchOrMismatch | ( | const CigarOperator & | op | ) | [inline, static] |
Return true if the specified operation is a match/mismatch operation, false if not.
Definition at line 313 of file Cigar.h.
References isMatchOrMismatch().
{ return(isMatchOrMismatch(op.operation)); }
bool Cigar::operator== | ( | Cigar & | rhs | ) | const |
const int32_t Cigar::INDEX_NA = -1 [static] |
Value associated with an index that is not applicable/does not exist, used for converting between query and reference indexes/offsets when an associated index/offset does not exist.
Definition at line 492 of file Cigar.h.
Referenced by PileupElementBaseQual::addEntry(), SamRecordHelper::checkSequence(), SamTags::createMDTag(), getExpandedCigarIndexFromQueryIndex(), getExpandedCigarIndexFromRefOffset(), SamQuerySeqWithRefIter::getNextMatchMismatch(), getQueryIndex(), getRefOffset(), getRefPosition(), SamQuerySeqWithRef::seqWithEquals(), SamQuerySeqWithRef::seqWithoutEquals(), and SamFilter::softClip().