// ***************************************************************************
// BamAlignment.h (c) 2009 Derek Barnett
// Marth Lab, Department of Biology, Boston College
-// All rights reserved.
// ---------------------------------------------------------------------------
-// Last modified: 19 April 2011 (DB)
+// Last modified: 7 October 2011 (DB)
// ---------------------------------------------------------------------------
// Provides the BamAlignment data structure
// ***************************************************************************
#include <api/api_global.h>
#include <api/BamAux.h>
+#include <api/BamConstants.h>
+#include <cstdlib>
+#include <cstring>
#include <string>
#include <vector>
// tag data access methods
public:
- // -------------------------------------------------------------------------------------
- // N.B. - The following tag access methods may not be used on BamAlignments fetched
- // using BamReader::GetNextAlignmentCore(). Attempting to use them will not result in
- // error message (to keep output clean) but will ALWAYS return false. Only user-created
- // BamAlignments or those retrieved using BamReader::GetNextAlignment() are valid here.
- //
- // You can call BuildCharData() on such an alignment retrieved by GetNextAlignmentCore().
- // This populates all the character data, and will enable subsequent queries on tag data.
- // -------------------------------------------------------------------------------------
-
- // adds a tag
- bool AddTag(const std::string& tag, const std::string& type, const std::string& value);
- bool AddTag(const std::string& tag, const std::string& type, const uint32_t& value);
- bool AddTag(const std::string& tag, const std::string& type, const int32_t& value);
- bool AddTag(const std::string& tag, const std::string& type, const float& value);
-
- // adds a "binary array" tag
- bool AddTag(const std::string& tag, const std::vector<uint8_t>& values);
- bool AddTag(const std::string& tag, const std::vector<int8_t>& values);
- bool AddTag(const std::string& tag, const std::vector<uint16_t>& values);
- bool AddTag(const std::string& tag, const std::vector<int16_t>& values);
- bool AddTag(const std::string& tag, const std::vector<uint32_t>& values);
- bool AddTag(const std::string& tag, const std::vector<int32_t>& values);
- bool AddTag(const std::string& tag, const std::vector<float>& values);
-
- // edits a tag
- bool EditTag(const std::string& tag, const std::string& type, const std::string& value);
- bool EditTag(const std::string& tag, const std::string& type, const uint32_t& value);
- bool EditTag(const std::string& tag, const std::string& type, const int32_t& value);
- bool EditTag(const std::string& tag, const std::string& type, const float& value);
-
- // edits a "binary array" tag
- bool EditTag(const std::string& tag, const std::vector<uint8_t>& values);
- bool EditTag(const std::string& tag, const std::vector<int8_t>& values);
- bool EditTag(const std::string& tag, const std::vector<uint16_t>& values);
- bool EditTag(const std::string& tag, const std::vector<int16_t>& values);
- bool EditTag(const std::string& tag, const std::vector<uint32_t>& values);
- bool EditTag(const std::string& tag, const std::vector<int32_t>& values);
- bool EditTag(const std::string& tag, const std::vector<float>& values);
-
- // retrieves data for a tag
- bool GetTag(const std::string& tag, std::string& destination) const;
- bool GetTag(const std::string& tag, uint32_t& destination) const;
- bool GetTag(const std::string& tag, int32_t& destination) const;
- bool GetTag(const std::string& tag, float& destination) const;
-
- // retrieves data for a "binary array" tag
- bool GetTag(const std::string& tag, std::vector<uint32_t>& destination) const;
- bool GetTag(const std::string& tag, std::vector<int32_t>& destination) const;
- bool GetTag(const std::string& tag, std::vector<float>& destination) const;
-
- // retrieves the BAM tag-type character for a tag
+ // add a new tag
+ template<typename T> bool AddTag(const std::string& tag, const std::string& type, const T& value);
+ template<typename T> bool AddTag(const std::string& tag, const std::vector<T>& values);
+
+ // edit (or append) tag
+ template<typename T> bool EditTag(const std::string& tag, const std::string& type, const T& value);
+ template<typename T> bool EditTag(const std::string& tag, const std::vector<T>& values);
+
+ // retrieves tag data
+ template<typename T> bool GetTag(const std::string& tag, T& destination) const;
+ template<typename T> bool GetTag(const std::string& tag, std::vector<T>& destination) const;
+
+ // retrieves the BAM type-code for requested tag
+ // (returns whether or not tag exists, and type-code is valid)
bool GetTagType(const std::string& tag, char& type) const;
// legacy methods (consider deprecated, but still available)
bool GetEditDistance(uint32_t& editDistance) const; // retrieves value of "NM" tag
bool GetReadGroup(std::string& readGroup) const; // retrieves value of "RG" tag
-
+
// returns true if alignment has a record for this tag name
bool HasTag(const std::string& tag) const;
// removes a tag
- bool RemoveTag(const std::string& tag);
+ void RemoveTag(const std::string& tag);
// additional methods
public:
// populates alignment string fields
bool BuildCharData(void);
+
// calculates alignment end position
int GetEndPosition(bool usePadded = false, bool zeroBased = true) const;
+ // returns a description of the last error that occurred
+ std::string GetErrorString(void) const;
+
// public data fields
public:
std::string Name; // read name
int32_t InsertSize; // mate-pair insert size
std::string Filename; // name of BAM file which this alignment comes from
+ //! \cond
+ // internal utility methods
+ private:
+ bool FindTag(const std::string& tag,
+ char*& pTagData,
+ const unsigned int& tagDataLength,
+ unsigned int& numBytesParsed) const;
+ bool IsValidSize(const std::string& tag,
+ const std::string& type) const;
+ void SetErrorString(const std::string& where,
+ const std::string& what) const;
+ bool SkipToNextTag(const char storageType,
+ char*& pTagData,
+ unsigned int& numBytesParsed) const;
+
// internal data
private:
- //! \cond
+
struct BamAlignmentSupportData {
// data members
BamAlignmentSupportData SupportData;
friend class Internal::BamReaderPrivate;
friend class Internal::BamWriterPrivate;
- //! \endcond
+
+ mutable std::string ErrorString; // mutable to allow updates even in logically const methods
+ //! \endcond
};
+// ---------------------------------------------------------
+// BamAlignment tag access methods
+
+template<typename T>
+inline bool BamAlignment::AddTag(const std::string& tag,
+ const std::string& type,
+ const T& value)
+{
+ // if char data not populated, do that first
+ if ( SupportData.HasCoreOnly )
+ BuildCharData();
+
+ // check tag/type size
+ if ( !IsValidSize(tag, type) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // check that storage type code is OK for T
+ if ( !TagTypeHelper<T>::CanConvertTo(type.at(0)) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // localize the tag data
+ char* pTagData = (char*)TagData.data();
+ const unsigned int tagDataLength = TagData.size();
+ unsigned int numBytesParsed = 0;
+
+ // if tag already exists, return false
+ // use EditTag explicitly instead
+ if ( FindTag(tag, pTagData, tagDataLength, numBytesParsed) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // otherwise, convert value to string
+ union { T value; char valueBuffer[sizeof(T)]; } un;
+ un.value = value;
+
+ // copy original tag data to temp buffer
+ const std::string newTag = tag + type;
+ const size_t newTagDataLength = tagDataLength + newTag.size() + sizeof(T); // leave room for new T
+ RaiiBuffer originalTagData(newTagDataLength);
+ memcpy(originalTagData.Buffer, TagData.c_str(), tagDataLength + 1); // '+1' for TagData null-term
+
+ // append newTag
+ strcat(originalTagData.Buffer + tagDataLength, newTag.data());
+ memcpy(originalTagData.Buffer + tagDataLength + newTag.size(), un.valueBuffer, sizeof(T));
+
+ // store temp buffer back in TagData
+ const char* newTagData = (const char*)originalTagData.Buffer;
+ TagData.assign(newTagData, newTagDataLength);
+ return true;
+}
+
+template<>
+inline bool BamAlignment::AddTag<std::string>(const std::string& tag,
+ const std::string& type,
+ const std::string& value)
+{
+ // if char data not populated, do that first
+ if ( SupportData.HasCoreOnly )
+ BuildCharData();
+
+ // check tag/type size
+ if ( !IsValidSize(tag, type) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // check that storage type code is OK for string
+ if ( !TagTypeHelper<std::string>::CanConvertTo(type.at(0)) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // localize the tag data
+ char* pTagData = (char*)TagData.data();
+ const unsigned int tagDataLength = TagData.size();
+ unsigned int numBytesParsed = 0;
+
+ // if tag already exists, return false
+ // use EditTag explicitly instead
+ if ( FindTag(tag, pTagData, tagDataLength, numBytesParsed) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // otherwise, copy tag data to temp buffer
+ const std::string newTag = tag + type + value;
+ const size_t newTagDataLength = tagDataLength + newTag.size() + 1; // leave room for null-term
+ RaiiBuffer originalTagData(newTagDataLength);
+ memcpy(originalTagData.Buffer, TagData.c_str(), tagDataLength + 1); // '+1' for TagData null-term
+
+ // append newTag (removes original null-term, then appends newTag + null-term)
+ strcat(originalTagData.Buffer + tagDataLength, newTag.data());
+
+ // store temp buffer back in TagData
+ const char* newTagData = (const char*)originalTagData.Buffer;
+ TagData.assign(newTagData, newTagDataLength);
+ return true;
+}
+
+template<typename T>
+inline bool BamAlignment::AddTag(const std::string& tag,
+ const std::vector<T>& values)
+{
+ // if char data not populated, do that first
+ if ( SupportData.HasCoreOnly )
+ BuildCharData();
+
+ // check for valid tag name length
+ if ( tag.size() != Constants::BAM_TAG_TAGSIZE )
+ return false;
+
+ // localize the tag data
+ char* pTagData = (char*)TagData.data();
+ const unsigned int tagDataLength = TagData.size();
+ unsigned int numBytesParsed = 0;
+
+ // if tag already exists, return false
+ // use EditTag explicitly instead
+ if ( FindTag(tag, pTagData, tagDataLength, numBytesParsed) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // build new tag's base information
+ char newTagBase[Constants::BAM_TAG_ARRAYBASE_SIZE];
+ memcpy( newTagBase, tag.c_str(), Constants::BAM_TAG_TAGSIZE );
+ newTagBase[2] = Constants::BAM_TAG_TYPE_ARRAY;
+ newTagBase[3] = TagTypeHelper<T>::TypeCode();
+
+ // add number of array elements to newTagBase
+ const int32_t numElements = values.size();
+ memcpy(newTagBase + 4, &numElements, sizeof(int32_t));
+
+ // copy current TagData string to temp buffer, leaving room for new tag's contents
+ const size_t newTagDataLength = tagDataLength +
+ Constants::BAM_TAG_ARRAYBASE_SIZE +
+ numElements*sizeof(T);
+ RaiiBuffer originalTagData(newTagDataLength);
+ memcpy(originalTagData.Buffer, TagData.c_str(), tagDataLength+1); // '+1' for TagData's null-term
+
+ // write newTagBase (removes old null term)
+ strcat(originalTagData.Buffer + tagDataLength, (const char*)newTagBase);
+
+ // add vector elements to tag
+ int elementsBeginOffset = tagDataLength + Constants::BAM_TAG_ARRAYBASE_SIZE;
+ for ( int i = 0 ; i < numElements; ++i ) {
+ const T& value = values.at(i);
+ memcpy(originalTagData.Buffer + elementsBeginOffset + i*sizeof(T), &value, sizeof(T));
+ }
+
+ // store temp buffer back in TagData
+ const char* newTagData = (const char*)originalTagData.Buffer;
+ TagData.assign(newTagData, newTagDataLength);
+ return true;
+}
+
+template<typename T>
+inline bool BamAlignment::EditTag(const std::string& tag,
+ const std::string& type,
+ const T& value)
+{
+ // if char data not populated, do that first
+ if ( SupportData.HasCoreOnly )
+ BuildCharData();
+
+ // remove existing tag if present, then append tag with new value
+ if ( HasTag(tag) )
+ RemoveTag(tag);
+ return AddTag(tag, type, value);
+}
+
+template<typename T>
+inline bool BamAlignment::EditTag(const std::string& tag,
+ const std::vector<T>& values)
+{
+ // if char data not populated, do that first
+ if ( SupportData.HasCoreOnly )
+ BuildCharData();
+
+ // remove existing tag if present, then append tag with new values
+ if ( HasTag(tag) )
+ RemoveTag(tag);
+ return AddTag(tag, values);
+}
+
+template<typename T>
+inline bool BamAlignment::GetTag(const std::string& tag,
+ T& destination) const
+{
+ // skip if alignment is core-only
+ if ( SupportData.HasCoreOnly ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // skip if no tags present
+ if ( TagData.empty() ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // localize the tag data
+ char* pTagData = (char*)TagData.data();
+ const unsigned int tagDataLength = TagData.size();
+ unsigned int numBytesParsed = 0;
+
+ // return failure if tag not found
+ if ( FindTag(tag, pTagData, tagDataLength, numBytesParsed) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // fetch data type
+ const char type = *(pTagData - 1);
+ if ( !TagTypeHelper<T>::CanConvertFrom(type) ) {
+ // TODO: set error string ?
+ return false;
+ }
+
+ // determine data length
+ int destinationLength = 0;
+ switch ( type ) {
+
+ // 1 byte data
+ case (Constants::BAM_TAG_TYPE_ASCII) :
+ case (Constants::BAM_TAG_TYPE_INT8) :
+ case (Constants::BAM_TAG_TYPE_UINT8) :
+ destinationLength = 1;
+ break;
+
+ // 2 byte data
+ case (Constants::BAM_TAG_TYPE_INT16) :
+ case (Constants::BAM_TAG_TYPE_UINT16) :
+ destinationLength = 2;
+ break;
+
+ // 4 byte data
+ case (Constants::BAM_TAG_TYPE_INT32) :
+ case (Constants::BAM_TAG_TYPE_UINT32) :
+ case (Constants::BAM_TAG_TYPE_FLOAT) :
+ destinationLength = 4;
+ break;
+
+ // var-length types not supported for numeric destination
+ case (Constants::BAM_TAG_TYPE_STRING) :
+ case (Constants::BAM_TAG_TYPE_HEX) :
+ case (Constants::BAM_TAG_TYPE_ARRAY) :
+ SetErrorString("BamAlignment::GetTag",
+ "cannot store variable length tag data into a numeric destination");
+ return false;
+
+ // unrecognized tag type
+ default:
+ const std::string message = std::string("invalid tag type: ") + type;
+ SetErrorString("BamAlignment::GetTag", message);
+ return false;
+ }
+
+ // store data in destination
+ destination = 0;
+ memcpy(&destination, pTagData, destinationLength);
+
+ // return success
+ return true;
+}
+
+template<>
+inline bool BamAlignment::GetTag<std::string>(const std::string& tag,
+ std::string& destination) const
+{
+ // skip if alignment is core-only
+ if ( SupportData.HasCoreOnly ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // skip if no tags present
+ if ( TagData.empty() ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // localize the tag data
+ char* pTagData = (char*)TagData.data();
+ const unsigned int tagDataLength = TagData.size();
+ unsigned int numBytesParsed = 0;
+
+ // return failure if tag not found
+ if ( FindTag(tag, pTagData, tagDataLength, numBytesParsed) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // otherwise copy data into destination
+ const unsigned int dataLength = strlen(pTagData);
+ destination.clear();
+ destination.resize(dataLength);
+ memcpy( (char*)destination.data(), pTagData, dataLength );
+
+ // return success
+ return true;
+}
+
+// retrieves "binary-array" tag data
+template<typename T>
+inline bool BamAlignment::GetTag(const std::string& tag,
+ std::vector<T>& destination) const
+{
+ // skip if alignment is core-only
+ if ( SupportData.HasCoreOnly ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // skip if no tags present
+ if ( TagData.empty() ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // localize the tag data
+ char* pTagData = (char*)TagData.data();
+ const unsigned int tagDataLength = TagData.size();
+ unsigned int numBytesParsed = 0;
+
+ // return false if tag not found
+ if ( !FindTag(tag, pTagData, tagDataLength, numBytesParsed) ) {
+ // TODO: set error string?
+ return false;
+ }
+
+ // check that tag is array type
+ const char tagType = *(pTagData - 1);
+ if ( tagType != Constants::BAM_TAG_TYPE_ARRAY ) {
+ SetErrorString("BamAlignment::GetTag", "cannot store a non-array tag in array destination");
+ return false;
+ }
+
+ // fetch element type
+ const char elementType = *pTagData;
+ if ( !TagTypeHelper<T>::CanConvertFrom(elementType) ) {
+ // TODO: set error string ?
+ return false;
+ }
+ ++pTagData;
+
+ // calculate length of each element in tag's array
+ int elementLength = 0;
+ switch ( elementType ) {
+ case (Constants::BAM_TAG_TYPE_ASCII) :
+ case (Constants::BAM_TAG_TYPE_INT8) :
+ case (Constants::BAM_TAG_TYPE_UINT8) :
+ elementLength = sizeof(uint8_t);
+ break;
+
+ case (Constants::BAM_TAG_TYPE_INT16) :
+ case (Constants::BAM_TAG_TYPE_UINT16) :
+ elementLength = sizeof(uint16_t);
+ break;
+
+ case (Constants::BAM_TAG_TYPE_INT32) :
+ case (Constants::BAM_TAG_TYPE_UINT32) :
+ case (Constants::BAM_TAG_TYPE_FLOAT) :
+ elementLength = sizeof(uint32_t);
+ break;
+
+ // var-length types not supported for numeric destination
+ case (Constants::BAM_TAG_TYPE_STRING) :
+ case (Constants::BAM_TAG_TYPE_HEX) :
+ case (Constants::BAM_TAG_TYPE_ARRAY) :
+ SetErrorString("BamAlignment::GetTag",
+ "invalid array data, variable-length elements are not allowed");
+ return false;
+
+ // unknown tag type
+ default:
+ const std::string message = std::string("invalid array element type: ") + elementType;
+ SetErrorString("BamAlignment::GetTag", message);
+ return false;
+ }
+
+ // get number of elements
+ int32_t numElements;
+ memcpy(&numElements, pTagData, sizeof(int32_t));
+ pTagData += 4;
+ destination.clear();
+ destination.reserve(numElements);
+
+ // read in elements
+ T value;
+ for ( int i = 0 ; i < numElements; ++i ) {
+ memcpy(&value, pTagData, sizeof(T));
+ pTagData += sizeof(T);
+ destination.push_back(value);
+ }
+
+ // return success
+ return true;
+}
+
typedef std::vector<BamAlignment> BamAlignmentVector;
} // namespace BamTools
projects / bamtools.git / blobdiff