Playlist(int count, const char *const aSongs[]) : mStream(0), mSource(0)
{
for (int i = 0; i < count; ++i) {
mSongs.push_back(aSongs[i]);
}
mCurrentSong = mSongs.begin();
Init();
}
// correct MD sepearte list
// also collect deletion base list
string correctMDstring(numList MDnum, stringList MDLetter, stringList &deletions)
{
string MDList;
int i = 0;
//make deletion list
for (int i = 0; i < MDnum.size(); i ++)
{
if (MDLetter[i][0] == '^')
{
deletions.push_back(MDLetter[i]);
}
}
MDList = creatMDstring(MDnum,MDLetter);
return MDList;
}
//Split states into individual state strings
bool PWM::_splitPositions(std::string& txt ,stringList& sts){
size_t start=0;
size_t end=0;
while(start!=std::string::npos){
end=txt.find("NAME:",start+1);
std::string st = txt.substr(start,end-start);
clear_whitespace(st, "#><");
sts.push_back(st);
start=txt.find("NAME:",end);
}
return true;
}
//Get all Name in order
bool PWM::_getOrderedPositionNames(stringList& pos, stringList& names){
for(size_t i=0;i<pos.size();i++){
size_t nameHeader=pos[i].find("NAME:");
if (nameHeader == std::string::npos){
continue;
}
size_t nameLineEnding=pos[i].find_first_of("\n",nameHeader);
std::string name = pos[i].substr(nameHeader+5,nameLineEnding-(nameHeader+5));
clear_whitespace(name, " \t\n");
if (names.contains(name)){
std::cerr << "Position with name of: " << name << " is defined twice in the model\n";
return false;
}
else{
names.push_back(name);
}
}
return true;
}
// Fix sequence, MDline, cigarLine
// i. collect clipped bases
// ii. collect insertion bases
// iii. remove insertion bases on
// 1. query sequence and
// 2. cigar line
// iv. count mismatch pattern (e.g. AtoG, GtoT)
stringList insertionAndMismatch(string cigarLine, string MDline, string sequence, stringList &insertion,
string &softclippedHead, string &softclippedTail, string id, int numOfMis)
{
stringList mismatchList(numOfMis);
string fixedCigarLine, fixedSeq, mismatch, ref, read;
int mis, i = 0, size = cigarLine.length();
while( i < size)
{
if (cigarLine.at(i) == 'S')
{
if (i == 0)
{
while ( cigarLine.at(i) == 'S')
{
i ++ ;
}
softclippedHead = sequence.substr(0,i);
}
else
{
softclippedTail = sequence.substr(i,size-i);
i += size;
}
}
else if (cigarLine.at(i) == 'I')
{
insertion.push_back(string(1,sequence[i]));
i ++;
}
else
{
fixedCigarLine.push_back(cigarLine[i]);
fixedSeq.push_back(sequence[i]);
i ++;
}
}
/*
//debugging
cerr << fixedCigarLine << endl;
cerr << MDline << endl;
cerr << fixedSeq <<endl;
cerr << sequence << endl ;
cerr << cigarLine << endl;
cerr << softclippedHead << '\t' << softclippedTail << endl;
*/
//assertion
assert(fixedCigarLine.length() == MDline.length());
assert(fixedSeq.length() == MDline.length());
//extract misincorporations
mis = 0;
for (int i = 0; i < fixedCigarLine.length(); i++)
{
if (fixedCigarLine.at(i) == 'M' && MDline.at(i) != '=')
{
ref = MDline.substr(i,1) ;
read = fixedSeq.substr(i,1);
assert (ref != read);
mismatch = ref +"to"+ read;
mismatchList[mis] = mismatch;
mis ++;
}
}
return mismatchList;
}