// converts a space separated quality string into a compressed quality string
// NOTE: this function has horrible amounts of overhead, but lean and mean code that I had before
// failed some of the unit tests.
void CRegexUtilities::ConvertQualities(string& qualities, CMosaikString& compQualities) {
string::iterator strIte = qualities.end() - 1;
while ( *strIte == ' ' ) {
qualities.erase( strIte );
strIte--;
}
vector<string> columns;
vector<string>::const_iterator sIter;
char* pQualities = (char*)qualities.c_str();
Chomp(pQualities);
back_insert_iterator<vector<string> > backiter(columns);
SplitString(backiter, " ", pQualities);
const unsigned int numQualities = (unsigned int)columns.size();
compQualities.Reserve(numQualities);
compQualities.SetLength(numQualities);
unsigned char* pCompQualities = (unsigned char*)compQualities.Data();
for(sIter = columns.begin(); sIter != columns.end(); ++sIter, ++pCompQualities) {
if(sIter->empty()) continue;
*pCompQualities = GetUnsignedChar((char*)sIter->c_str());
}
}
// converts the supplied read from pseudo-colorspace to colorspace
// The function is used by the unaligned-read writer.
void CColorspaceUtilities::ConvertReadPseudoColorspaceToColorspace(CMosaikString& s) {
char* pBases = s.Data();
for(unsigned int i = 0; i < s.Length(); ++i, ++pBases) {
switch(*pBases) {
case 'A':
*pBases = '0';
break;
case 'C':
*pBases = '1';
break;
case 'G':
*pBases = '2';
break;
case 'T':
*pBases = '3';
break;
case 'X':
case 'N':
break;
default:
printf("ERROR: Unrecognized nucleotide (%c) when converting read to colorspace.\n", pBases[i]);
exit(1);
break;
}
}
}
// converts the supplied read from colorspace to pseudo-colorspace
void CColorspaceUtilities::ConvertReadColorspaceToPseudoColorspace(CMosaikString& s) {
char* pBases = s.Data();
for(unsigned int i = 0; i < s.Length(); ++i, ++pBases) {
switch(*pBases) {
case '0':
*pBases = 'A';
break;
case '1':
*pBases = 'C';
break;
case '2':
*pBases = 'G';
break;
case '3':
*pBases = 'T';
break;
case 'X':
break;
case '-':
*pBases = 'N';
break;
case '.':
// here we pick an arbitrary colorspace transition, this will have at
// least 25 % of being correct as opposed to specifying an 'N'.
*pBases = 'A';
break;
default:
printf("ERROR: Unrecognized nucleotide (%c) when converting read to pseudo-colorspace.\n", pBases[i]);
exit(1);
break;
}
}
}
// converts the supplied read from basespace to pseudo-colorspace
void CColorspaceUtilities::ConvertReadBasespaceToPseudoColorspace(CMosaikString& s) {
char* pPrev = s.Data();
char* pString = pPrev + 1;
// simplify various ambiguity codes
*pPrev = GetSimplifiedBase(*pPrev);
CS_MAP_t::const_iterator csIter;
for(unsigned int i = 1; i < s.Length(); ++i, ++pString, ++pPrev) {
// simplify various ambiguity codes
*pString = GetSimplifiedBase(*pString);
csIter = mCSMap.find(PACK_SHORT(*pPrev, *pString));
if(csIter == mCSMap.end()) {
printf("ERROR: Unknown combination found when converting to colorspace: [%c] & [%c]\n", *pPrev, *pString);
exit(1);
}
*pPrev = csIter->second;
}
// adjust the read
s.TrimEnd(1);
}
// encodes the supplied query sequence into 4-bit notation
void CBamWriter::EncodeQuerySequence(const CMosaikString& query, string& encodedQuery) {
// prepare the encoded query string
const unsigned int queryLen = query.Length();
const unsigned int encodedQueryLen = (unsigned int)((queryLen / 2.0) + 0.5);
encodedQuery.resize(encodedQueryLen);
char* pEncodedQuery = (char*)encodedQuery.data();
const char* pQuery = (const char*)query.CData();
unsigned char nucleotideCode;
bool useHighWord = true;
while(*pQuery) {
switch(*pQuery) {
case '=':
nucleotideCode = 0;
break;
case 'A':
nucleotideCode = 1;
break;
case 'C':
nucleotideCode = 2;
break;
case 'G':
nucleotideCode = 4;
break;
case 'T':
nucleotideCode = 8;
break;
case 'N':
nucleotideCode = 15;
break;
default:
printf("ERROR: Only the following bases are supported in the BAM format: {=, A, C, G, T, N}. Found [%c]\n", *pQuery);
exit(1);
}
// pack the nucleotide code
if(useHighWord) {
*pEncodedQuery = nucleotideCode << 4;
useHighWord = false;
} else {
*pEncodedQuery |= nucleotideCode;
++pEncodedQuery;
useHighWord = true;
}
// increment the query position
++pQuery;
}
}
// extracts the genome assembly ID from a FASTA/FASTQ header
void CRegexUtilities::ExtractGenomeAssemblyID(const string& line, CMosaikString& genomeAssemblyID) {
#ifdef WIN32
cmatch results;
if(!regex_search(line.c_str(), results, mGenomeAssemblyIDRegex)) {
genomeAssemblyID.SetLength(0);
return;
}
genomeAssemblyID = results[1].str().c_str();
#else
// TODO: replace this with the TR1 regex above when it finally works in gcc. It doesn't work in gcc 4.3.3
// find the GA tag
const string gaTag = "GA(";
string::size_type gaPos = line.find(gaTag.c_str());
if(gaPos == string::npos) {
genomeAssemblyID.SetLength(0);
return;
}
// find the matching end parenthesis
const unsigned int start = gaPos + gaTag.size();
unsigned int stop = start;
const char* pBuffer = line.data();
unsigned int lineLen = line.size();
if(stop < lineLen) {
while(pBuffer[stop] != ')') {
stop++;
if(stop == lineLen) break;
}
}
if(start == stop) {
cout << "ERROR: could not parse genome assembly ID from FASTA header." << endl;
cout << " " << line << endl;
exit(1);
}
genomeAssemblyID = line.substr(start, stop - start).c_str();
#endif
}
// load the read header from disk
void CAlignmentReader::LoadReadHeader(
CMosaikString& readName,
unsigned int& readGroupCode,
unsigned char& readStatus,
int& numMate1Alignments,
int& numMate2Alignments,
int& numMate1OriginalAlignments,
int& numMate2OriginalAlignments,
int& numMate1Hashes,
int& numMate2Hashes) {
// get the read name
const unsigned char readNameLength = (unsigned char)*mBufferPtr;
++mBufferPtr;
readName.Copy((const char*)mBufferPtr, readNameLength);
mBufferPtr += readNameLength;
// get the read group code
memcpy((char*)&readGroupCode, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
// get the read status
readStatus = (unsigned char)*mBufferPtr;
++mBufferPtr;
const bool haveMate1 = ((readStatus & RF_HAVE_MATE1) != 0 ? true : false);
const bool haveMate2 = ((readStatus & RF_HAVE_MATE2) != 0 ? true : false);
// get the number of mate 1 alignments
if(haveMate1) {
memcpy((char*)&numMate1Alignments, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
memcpy((char*)&numMate1OriginalAlignments, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
memcpy((char*)&numMate1Hashes, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
}
// get the number of mate 2 alignments
if(haveMate2) {
memcpy((char*)&numMate2Alignments, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
memcpy((char*)&numMate2OriginalAlignments, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
memcpy((char*)&numMate2Hashes, mBufferPtr, SIZEOF_INT);
mBufferPtr += SIZEOF_INT;
}
}
// saves the alignment to the alignment archive
void CBamWriter::SaveAlignment(
const Alignment& al,
const char* zaString,
const bool& noCigarMdNm,
const bool& notShowRnamePos,
const bool& isSolid,
const bool& processedBamData,
const bool& report_zn) {
// =================
// set the BAM flags
// =================
// define our flags
unsigned int flag = 0;
int insertSize = 0;
if(al.IsPairedEnd) {
flag |= BAM_SEQUENCED_AS_PAIRS;
// first or second mate?
flag |= (al.IsFirstMate ? BAM_QUERY_FIRST_MATE : BAM_QUERY_SECOND_MATE);
if(al.IsResolvedAsPair) {
if ( al.IsResolvedAsProperPair )
flag |= BAM_PROPER_PAIR;
if(al.IsMateMapped && al.IsMateReverseStrand) flag |= BAM_MATE_REVERSE_COMPLEMENT;
insertSize = al.FragmentLength;
}
if ( !al.IsMapped )
flag |= BAM_QUERY_UNMAPPED;
if ( !al.IsMateMapped )
flag |= BAM_MATE_UNMAPPED;
} else {
if ( !al.IsMapped )
flag |= BAM_QUERY_UNMAPPED;
}
if(al.IsMapped && al.IsReverseStrand) flag |= BAM_QUERY_REVERSE_COMPLEMENT;
// ==========================
// construct the cigar string
// ==========================
string packedCigar;
unsigned short numCigarOperations = 0;
if ( !noCigarMdNm ) {
if ( !processedBamData )
CreatePackedCigar( al, packedCigar, numCigarOperations, isSolid );
else {
packedCigar = al.PackedCigar;
numCigarOperations = al.NumCigarOperation;
}
}
else
packedCigar = "\0";
const unsigned int packedCigarLen = !noCigarMdNm ? packedCigar.size() : 0;
// ===================
// write the alignment
// ===================
// remove the gaps from the read
CMosaikString query;
if ( !processedBamData ) {
query = al.Query.CData();
query.Remove('-');
}
// initialize
const unsigned int nameLen = al.Name.Length() + 1;
const unsigned int queryLen = processedBamData ? al.QueryLength : query.Length();
// sanity check
//al.BaseQualities.CheckQuality();
//if ( queryLen != alIter->BaseQualities.Length() ) {
// printf("ERROR: The lengths of bases(%u) and qualities(%u) of Read (%s) didn't match.\n", queryLen, alIter->BaseQualities.Length(), readName.CData());
// exit(1);
//}
// encode the query
string encodedQuery;
if (!processedBamData && (query.Length() != 0))
EncodeQuerySequence(query, encodedQuery);
else
encodedQuery = al.EncodedQuery;
const unsigned int encodedQueryLen = encodedQuery.size();
// create our read group tag
string readGroupTag;
const unsigned int readGroupTagLen = 3 + al.ReadGroup.size() + 1;
readGroupTag.resize(readGroupTagLen);
char* pReadGroupTag = (char*)readGroupTag.data();
sprintf(pReadGroupTag, "RGZ%s", al.ReadGroup.c_str());
// create our mismatch tag
string mismatchTag;
unsigned int numMismatches = 0;
unsigned int nmTagLen = 0;
if ( !noCigarMdNm ) {
mismatchTag = "NMi";
mismatchTag.resize(MISMATCH_TAG_LEN);
nmTagLen = MISMATCH_TAG_LEN;
numMismatches = al.NumMismatches;
memcpy((char*)mismatchTag.data() + 3, (char*)&numMismatches, SIZEOF_INT);
}
// create our MD tag
string mdTag;
char* pMd = 0;
unsigned int mdTagLen = 0;
char* pMdTag;
if ( !noCigarMdNm ) {
if ( !processedBamData )
pMd = (char*) mdTager.GetMdTag( al.Reference.CData(), al.Query.CData(), al.Reference.Length() );
else
pMd = (char*) al.MdString.c_str();
mdTagLen = 3 + strlen( pMd ) + 1;
mdTag.resize( mdTagLen );
pMdTag = (char*)mdTag.data();
sprintf(pMdTag, "MDZ%s", pMd);
#ifdef VERBOSE_DEBUG
fprintf(stderr, "=== MD ===\n");
fprintf(stderr, "%s\n%s\n%s\n", al.Reference.CData(), al.Query.CData(), mdTag.c_str());
#endif
}
// create our za tag
unsigned int zaTagLen = 0;
string zaTag;
char* pZaTag;
if ((zaString != 0) && (zaString != (char)0)) {
zaTagLen = 3 + strlen( zaString ) + 1;
zaTag.resize( zaTagLen );
pZaTag = (char*)zaTag.data();
sprintf(pZaTag, "ZAZ%s",zaString);
}
// create our zn tag
unsigned int znTagLen = 0;
string znTag;
if (report_zn){
ostringstream zn_buffer;
zn_buffer << "ZNZ"
<< al.SwScore << ";"
<< al.NextSwScore << ";"
<< al.NumLongestMatchs << ";"
<< al.Entropy << ";"
<< al.NumMapped << ";"
<< al.NumHash;
znTag = zn_buffer.str();
znTagLen = znTag.size() + 1;
//cerr << znTag.data() << "\t" << znTag << "\t" << znTagLen << endl;
}
// create our cs tag
unsigned int csTagLen = 0;
string csTag;
char* pCsTag;
if (isSolid) {
csTagLen = 3 + strlen ( al.CsQuery.c_str() ) + 1;
csTag.resize( csTagLen );
pCsTag = (char*)csTag.data();
sprintf( pCsTag, "CSZ%s", al.CsQuery.c_str() );
}
// create our cq tag
unsigned int cqTagLen = 0;
string cqTag;
char* pCqTag;
if (isSolid) {
cqTagLen = 3 + strlen ( al.CsBaseQualities.c_str() ) + 1;
cqTag.resize( cqTagLen );
pCqTag = (char*)cqTag.data();
sprintf( pCqTag, "CQZ%s", al.CsBaseQualities.c_str() );
}
// retrieve our bin
unsigned int bin = CalculateMinimumBin(al.ReferenceBegin, al.ReferenceEnd);
// assign the BAM core data
unsigned int buffer[8] = {0};
unsigned int reference_index, reference_pos;
if (!al.IsMapped) {
if (al.IsMateMapped) reference_index = al.MateReferenceIndex;
else reference_index = 0xffffffff;
} else {
reference_index = al.ReferenceIndex;
}
if (!al.IsMapped) {
if (al.IsMateMapped) reference_pos = al.MateReferenceBegin;
else reference_pos = 0xffffffff;
} else {
reference_pos = al.ReferenceBegin;
}
buffer[0] = (notShowRnamePos) ? 0xffffffff : reference_index;
buffer[1] = (notShowRnamePos) ? 0xffffffff : reference_pos;
buffer[2] = (bin << 16) | (al.RecalibratedQuality << 8) | nameLen;
buffer[3] = (flag << 16) | numCigarOperations;
buffer[4] = queryLen;
if(al.IsPairedEnd) {
if (notShowRnamePos) {
buffer[5] = 0xffffffff;
buffer[6] = 0xffffffff;
} else {
if (!al.IsMateMapped) {//unmapped mate
buffer[5] = reference_index;
buffer[6] = reference_pos;
} else {
buffer[5] = al.MateReferenceIndex;
buffer[6] = al.MateReferenceBegin;
}
}
buffer[7] = insertSize;
} else {
buffer[5] = 0xffffffff;
buffer[6] = 0xffffffff;
buffer[7] = 0;
}
// write the block size
const unsigned int dataBlockSize = nameLen + packedCigarLen + encodedQueryLen + queryLen + readGroupTagLen + nmTagLen + mdTagLen + zaTagLen + znTagLen + csTagLen + cqTagLen;
const unsigned int blockSize = BAM_CORE_SIZE + dataBlockSize;
BgzfWrite((char*)&blockSize, SIZEOF_INT);
// write the BAM core
BgzfWrite((char*)&buffer, BAM_CORE_SIZE);
// write the query name
BgzfWrite(al.Name.CData(), nameLen);
// write the packed cigar
BgzfWrite(packedCigar.data(), packedCigarLen);
// write the encoded query sequence
BgzfWrite(encodedQuery.data(), encodedQueryLen);
// write the base qualities
BgzfWrite(al.BaseQualities.CData(), queryLen);
// write the read group tag
BgzfWrite(readGroupTag.data(), readGroupTagLen);
// write the mismatch tag
if ( !noCigarMdNm )
BgzfWrite(mismatchTag.data(), MISMATCH_TAG_LEN);
// write the MD tag
if ( !noCigarMdNm )
BgzfWrite(mdTag.data(), mdTagLen);
// write the ZA tag
if ( zaString != 0 && (zaString != (char)0))
BgzfWrite(zaTag.data(), zaTagLen);
// write the ZN tag
if (report_zn && (znTagLen > 0))
BgzfWrite(znTag.data(), znTagLen);
// write the cs tag
if (isSolid)
BgzfWrite(csTag.data(), csTagLen);
// write the cq tag
if (isSolid)
BgzfWrite(cqTag.data(), cqTagLen);
}
// converts the supplied alignment from colorspace to basespace
void CColorspaceUtilities::ConvertAlignmentToBasespace(Alignment& al) {
// convert the alignment to character arrays
const unsigned int pairwiseLen = al.Reference.Length();
//char* pReference = al.Reference.Data();
//char* pQuery = al.Query.Data();
// re-allocate mBsRef & mBsQuery if the reversed space is insufficient
if( pairwiseLen > mCsAl.csAlignmentLength ) {
//if ( mCsAl.csReference ) delete [] mCsAl.csReference;
//if ( mCsAl.csQuery ) delete [] mCsAl.csQuery;
//if ( mCsAl.bsReference ) delete [] mCsAl.bsReference;
//if ( mCsAl.bsQuery ) delete [] mCsAl.bsQuery;
//if ( mCsAl.type ) delete [] mCsAl.type;
//if ( mCsAl.dashReference ) delete [] mCsAl.dashReference;
//if ( mCsAl.dashQuery ) delete [] mCsAl.dashQuery;
//if ( mCsAl.mismatch ) delete [] mCsAl.mismatch;
//if ( mCsAl.identical ) delete [] mCsAl.identical;
delete [] mCsAl.csReference;
delete [] mCsAl.csQuery;
delete [] mCsAl.bsReference;
delete [] mCsAl.bsQuery;
delete [] mCsAl.type;
delete [] mCsAl.dashReference;
delete [] mCsAl.dashQuery;
delete [] mCsAl.mismatch;
delete [] mCsAl.identical;
try {
mCsAl.csAlignmentLength = pairwiseLen;
mCsAl.csReference = new char [ pairwiseLen ];
mCsAl.csQuery = new char [ pairwiseLen ];
mCsAl.bsReference = new char [ pairwiseLen + 2 ];
mCsAl.bsQuery = new char [ pairwiseLen + 2 ];
mCsAl.type = new unsigned short [pairwiseLen];
mCsAl.dashReference = new RegionT [ pairwiseLen ];
mCsAl.dashQuery = new RegionT [ pairwiseLen ];
mCsAl.mismatch = new unsigned int [ pairwiseLen ];
mCsAl.identical = new RegionT [ pairwiseLen ];
}
catch( bad_alloc ) {
cout << "ERROR: Unable to allocate enough memory for converting colorspace ." << endl;
exit(1);
}
}
// initialize the counters
mCsAl.nDashReference = 0;
mCsAl.nDashQuery = 0;
mCsAl.nMismatch = 0;
mCsAl.nIdentical = 0;
// convert cs to bs
// initial the first BS base
char bsBase = mpBsRefSeqs[al.ReferenceIndex][al.ReferenceBegin];
if ( bsBase == 'N' || bsBase == 'X' ) {
cout << "ERROR: The first base of the colorspace-basespace converter is N or X." << endl;
exit(1);
}
// copy CS alignments
memcpy ( mCsAl.csReference, al.Reference.Data(), pairwiseLen );
memcpy ( mCsAl.csQuery, al.Query.Data(), pairwiseLen );
mCsAl.bsReference[0] = bsBase;
mCsAl.bsQuery[0] = bsBase;
ConvertCs2Bs(mCsAl.csReference, mCsAl.bsReference, 0, pairwiseLen-1, bsBase);
ConvertCs2Bs(mCsAl.csQuery, mCsAl.bsQuery, 0, pairwiseLen-1, bsBase);
// search the dash regions & mismatches
bool continuedDReference = false;
bool continuedDQuery = false;
unsigned int nMatch = 0;
BS_MAP_t::const_iterator bsIter;
for ( unsigned int i = 0; i < pairwiseLen; i++ ) {
// determine identical region
const bool isEndIdentity = ( mCsAl.csQuery[i] != mCsAl.csReference[i] ) && ( nMatch >= mNAllowedMismatch );
if ( isEndIdentity ) {
mCsAl.identical[ mCsAl.nIdentical ].Begin = i - nMatch;
mCsAl.identical[ mCsAl.nIdentical ].Length = nMatch;
mCsAl.nIdentical++;
}
if ( mCsAl.csQuery[i] == mCsAl.csReference[i])
nMatch++;
else
nMatch = 0;
// determine mismatches
bool isN = (mCsAl.csReference[i] != 'A') && (mCsAl.csReference[i] != 'C') && (mCsAl.csReference[i] != 'G') && (mCsAl.csReference[i] != 'T');
const bool isMismatch = (mCsAl.csReference[i] != '-') && (mCsAl.csQuery[i] != '-') && !isN && (mCsAl.csReference[i] != mCsAl.csQuery[i]);
if ( isMismatch ) {
// set the position
mCsAl.mismatch[mCsAl.nMismatch] = i;
mCsAl.nMismatch++;
// set the mismatch flag
mCsAl.type[i] = 0;
}
// for reference convertion
if ( mCsAl.csReference[i] != '-' ) {
// end the current dash region
if ( continuedDReference ) {
mCsAl.nDashReference++;
// the current position could be a mismatch
mCsAl.mismatch[mCsAl.nMismatch] = i;
mCsAl.nMismatch++;
mCsAl.type[i] = 1;
}
continuedDReference = false;
}
else {
// start a dash region
if ( !continuedDReference ) {
mCsAl.dashReference[mCsAl.nDashReference].Begin = i;
mCsAl.dashReference[mCsAl.nDashReference].Length = 0;
// the preceding position could be a mismatch
mCsAl.mismatch[mCsAl.nMismatch] = i;
mCsAl.nMismatch++;
mCsAl.type[i] = 3;
}
mCsAl.dashReference[mCsAl.nDashReference].Length++;
continuedDReference = true;
}
// for query convertion
if ( mCsAl.csQuery[i] != '-' ) {
// end the current dash region
if ( continuedDQuery ) {
mCsAl.nDashQuery++;
// the current position could be a mismatch
mCsAl.mismatch[mCsAl.nMismatch] = i;
mCsAl.nMismatch++;
mCsAl.type[i] = 2;
}
continuedDQuery = false;
}
else {
// start a dash region
if ( !continuedDQuery ) {
mCsAl.dashQuery[mCsAl.nDashQuery].Begin = i;
mCsAl.dashQuery[mCsAl.nDashQuery].Length = 0;
// the preceding position could be a mismatch
mCsAl.mismatch[mCsAl.nMismatch] = i;
mCsAl.nMismatch++;
mCsAl.type[i] = 4;
}
mCsAl.dashQuery[mCsAl.nDashQuery].Length++;
continuedDQuery = true;
}
}
if ( nMatch > 0 ) {
mCsAl.identical[ mCsAl.nIdentical ].Begin = pairwiseLen - nMatch;
mCsAl.identical[ mCsAl.nIdentical ].Length = nMatch;
mCsAl.nIdentical++;
}
if ( mCsAl.identical[mCsAl.nIdentical - 1].Begin != 0 ) {
// find sequencing errors
if ( mCsAl.nMismatch > 0 )
FindSequencingError(pairwiseLen);
if ( mCsAl.nDashReference > 0 ) {
for ( unsigned int i = 0; i < mCsAl.nDashReference; i++ ) {
unsigned int curPosition = mCsAl.dashReference[i].Begin + mCsAl.dashReference[i].Length;
if ( mCsAl.bsReference[ curPosition ] != mCsAl.bsQuery[ curPosition ] ) {
curPosition = mCsAl.dashReference[i].Begin;
for ( unsigned int j = 0; j < mCsAl.dashReference[i].Length; j++ )
mCsAl.bsQuery[ curPosition + j + 1 ] = 'N';
}
}
AdjustDash(mCsAl.csReference, mCsAl.csQuery, mCsAl.dashReference, mCsAl.nDashReference, mCsAl.bsReference);
}
if ( mCsAl.nDashQuery > 0 )
AdjustDash(mCsAl.csQuery, mCsAl.csReference, mCsAl.dashQuery, mCsAl.nDashQuery, mCsAl.bsQuery);
// deal with the indentical region
for ( unsigned int i = 0; i < mCsAl.nIdentical; i++ ) {
unsigned int csEnd = mCsAl.identical[i].Begin + mCsAl.identical[i].Length - 1;
unsigned int curPosition = mCsAl.identical[i].Begin;
bsBase = mCsAl.bsReference[ curPosition ];
while ( ( bsBase == '-' ) || ( bsBase == 'N' ) ) {
curPosition--;
bsBase = mCsAl.bsReference[ curPosition ];
if ( curPosition == 0 )
break;
}
bool isGoodBsBase = false;
if ( ( bsBase != '-' ) && ( bsBase != 'N' ) )
isGoodBsBase = true;
if ( isGoodBsBase )
ConvertCs2Bs(mCsAl.csQuery, mCsAl.bsQuery, mCsAl.identical[i].Begin, csEnd, bsBase);
}
}
// end up the sequences
mCsAl.bsReference[ pairwiseLen + 1 ] = 0;
mCsAl.bsQuery[ pairwiseLen + 1 ] = 0;
al.Reference = mCsAl.bsReference;
al.Query = mCsAl.bsQuery;
++al.ReferenceEnd;
++al.QueryEnd;
al.QueryLength = al.QueryEnd - al.QueryBegin + 1;
// ------------------------------------------------------------------------------------------
// convert the colorspace transition qualities to base qualities
// NOTE: this algorithm will simply take the minimum of the two qualities that overlap a base
// ------------------------------------------------------------------------------------------
const unsigned short numColorspaceQualities = al.BaseQualities.Length();
const unsigned short lastCSQIndex = numColorspaceQualities - 1;
CMosaikString csQualities = al.BaseQualities;
al.BaseQualities.Reserve(numColorspaceQualities + 1);
al.BaseQualities.SetLength(numColorspaceQualities + 1);
const char* pCSQual = csQualities.CData();
char* pBSQual = al.BaseQualities.Data();
// handle the first base quality
*pBSQual = *pCSQual;
++pBSQual;
// handle the internal base qualities
for(unsigned short i = 1; i < numColorspaceQualities; ++i, ++pBSQual)
*pBSQual = min(pCSQual[i - 1], pCSQual[i]);
// handle the final base quality
*pBSQual = pCSQual[lastCSQIndex];
// update the number of mismatches
// TODO: This should be augmented to support IUPAC ambiguity codes
const unsigned int bsPairwiseLen = al.Reference.Length();
al.NumMismatches = 0;
for(unsigned short i = 0; i < bsPairwiseLen; ++i) {
if(mCsAl.bsReference[i] != mCsAl.bsQuery[i]) al.NumMismatches++;
}
}
