bool CSplignTrim::ThrowAway20_28_90(TSeg& s)
{
if( s.m_len < 20 || // 20 rule
( s.m_idty < 0.9 && s.m_len < 28 ) ) {// 28_90 rule
s.SetToGap();
return true;
}
return false;
}
//legacy check
//it two short throws away and returns true
//otherwise returns false
bool CSplignTrim::ThrowAwayShortExon(TSeg& s)
{
const size_t min_query_size = 4;
if( int(s.m_box[1] - s.m_box[0] + 1) < int(min_query_size)) {
s.SetToGap();
return true;
}
return false;
}
void CSplignTrim::CutToMatchLeft(TSeg& s)
{
size_t pos = s.m_details.find('M');
if(pos == string::npos) {
s.SetToGap();
return;
}
if(pos > 0) {
CutFromLeft(pos, s);
}
}
void CSplignTrim::CutToMatchRight(TSeg& s)
{
size_t pos = s.m_details.rfind('M');
if(pos == string::npos) {
s.SetToGap();
return;
}
size_t len = s.m_details.length() - pos - 1;//length to cut
if(len > 0) {
CutFromRight(len, s);
}
}
void CSplignTrim::CutFromRight(size_t len, TSeg& s) //len is length on alignment to cut
{
if(len == 0) return;
if(ThrowAwayShortExon(s)) return;
string::reverse_iterator irs, irs0, irs1;
irs0 = s.m_details.rbegin();
irs1 = s.m_details.rend();
if(irs1 - irs0 <= (int)len) {//cut all
s.SetToGap();
return;
}
irs1 = irs0 + len;
int i0 = 0, i1 = 0;
for(irs = irs0; irs != irs1; ++irs) {
switch(*irs) {
case 'M':
case 'R': {
++i0;
++i1;
}
break;
case 'I': {
++i1;
}
break;
case 'D': {
++i0;
}
}
}
// resize
s.m_box[1] -= i0;
if(ThrowAwayShortExon(s)) return;//too short
s.m_box[3] -= i1;
s.m_details.resize(s.m_details.size() - len);
s.Update(m_aligner.GetNonNullPointer());
// update the last two annotation chars
const size_t adim = s.m_annot.size();
if(adim > 2 && s.m_annot[adim - 3] == '>') {
const char c3 (s.m_box[3] + 1 < (size_t)m_seqlen? m_seq[s.m_box[3] + 1]: ' ');
const char c4 (s.m_box[3] + 2 < (size_t)m_seqlen? m_seq[s.m_box[3] + 2]: ' ');
s.m_annot[adim-2] = c3;
s.m_annot[adim-1] = c4;
}
}
void CSplignTrim::CutFromLeft(size_t len, TSeg& s) //len is length on alignment to cut
{
if(len == 0) return;
if(ThrowAwayShortExon(s)) return;
string::iterator irs, irs0, irs1;
irs0 = s.m_details.begin();
irs1 = s.m_details.end();
if(irs1 - irs0 <= (int)len) {//cut all
s.SetToGap();
return;
}
irs1 = irs0 + len;
int i0 = 0, i1 = 0;
for(irs = irs0; irs != irs1; ++irs) {
switch(*irs) {
case 'M':
case 'R': {
++i0;
++i1;
}
break;
case 'I': {
++i1;
}
break;
case 'D': {
++i0;
}
}
}
// resize
s.m_box[0] += i0;
if(ThrowAwayShortExon(s)) return;//too short
s.m_box[2] += i1;
s.m_details.erase(0, len);
s.Update(m_aligner.GetNonNullPointer());
// update the first two annotation symbols
if(s.m_annot.size() > 2 && s.m_annot[2] == '<') {
int j1 = int(s.m_box[2]) - 2;
char c1 = j1 >= 0? m_seq[j1]: ' ';
s.m_annot[0] = c1;
int j2 = int(s.m_box[2]) - 1;
char c2 = j2 >= 0? m_seq[j2]: ' ';
s.m_annot[1] = c2;
}
}