BedLine BedLine::fromString(QString str)
{
//normalize
str = str.replace(',', ""); //remove thousands separator
str = str.replace(':', '\t').replace('-', '\t'); //also accept "[c]:[s]-[e]"
str = str.replace(QRegExp("[ ]+"), "\t"); //also accept "[c] [s] [e]" (with any number of spaces)
//split
QStringList parts = str.split('\t');
if (parts.count()<3) return BedLine();
//convert
try
{
return BedLine(parts[0], Helper::toInt(parts[1]), Helper::toInt(parts[2]));
}
catch(...)
{
return BedLine();
}
}
void BedFile::load(QString filename)
{
clear();
//parse from stream
QSharedPointer<QFile> file = Helper::openFileForReading(filename, true);
while(!file->atEnd())
{
QByteArray line = file->readLine();
while (line.endsWith('\n') || line.endsWith('\r')) line.chop(1);
//skip empty lines and header lines
if(line.length()==0 || line.startsWith("track ") || line.startsWith("browser ")) continue;
//error when less than 3 fields
QList<QByteArray> fields = line.split('\t');
if (fields.count()<3)
{
THROW(FileParseException, "BED file line with less than three fields found: '" + line.trimmed() + "'");
}
//error when chromosome is empty
if (fields[0]=="")
{
THROW(FileParseException, "Empty BED file chromosome field '" + fields[0] + "'!");
}
//error on position converion
int start_pos = Helper::toInt(fields[1], "start position", line);
int end_pos = Helper::toInt(fields[2], "end position", line);
//create line
QStringList annos;
for (int i=3; i<fields.count(); ++i)
{
annos.append(fields[i]);
}
append(BedLine(fields[0], start_pos+1, end_pos, annos));
}
}
void GeneSelectorDialog::updateGeneTable()
{
//clear details
ui->details->clearContents();
//convert input to gene list
QStringList genes = NGSHelper::textToGenes(ui->genes->toPlainText());
if (genes.isEmpty()) return;
//set cursor
QApplication::setOverrideCursor(QCursor(Qt::BusyCursor));
ui->details->blockSignals(true); //otherwise itemChanged is emitted
//check for CNA results
QString folder = QFileInfo(bam_file_).absolutePath();
QStringList files = Helper::findFiles(folder, "*_cnvs.seg", false);
bool cna_result_present = (files.count()==1);
//load CNA results
BedFile cna_results;
if (cna_result_present)
{
auto f = Helper::openFileForReading(files[0]);
while(!f->atEnd())
{
QString line = f->readLine();
//skip headers
if (line.isEmpty() || line[0]!='\t')
{
continue;
}
//parse content
QStringList parts = line.split('\t');
if (parts.count()<6) THROW(FileParseException, "SEG file line invalid: " + line);
Chromosome chr(parts[1]);
int start = Helper::toInt(parts[2], "SEG start position", line);
int end = Helper::toInt(parts[3], "SEG end position", line);
cna_results.append(BedLine(chr, start, end, QStringList() << parts[5]));
}
}
//load low-coverage file for processing system
files = Helper::findFiles(folder, "*_lowcov.bed", false);
if(files.count()!=1)
{
updateError("Gene selection error", "Low-coverage BED file not found in " + folder);
return;
}
BedFile sys_gaps;
sys_gaps.load(files[0]);
//load processing system target region
NGSD db;
QString sys_file = db.getProcessingSystem(bam_file_, NGSD::FILE);
if (sys_file=="")
{
updateError("Gene selection error", "Processing system target region BED file not found in " + folder);
return;
}
BedFile sys_roi;
sys_roi.load(sys_file);
//display genes
ui->details->setRowCount(genes.count());
for (int r=0; r<genes.count(); ++r)
{
//convert gene to approved symbol
QString gene = genes[r];
int gene_id = db.geneToApprovedID(gene);
if(gene_id==-1)
{
updateError("Gene selection error", "Gene symbol '" + gene + "' is not an approved symbol!");
return;
}
gene = db.geneSymbol(gene_id);
setGeneTableItem(r, 0, gene, Qt::AlignLeft, Qt::ItemIsUserCheckable|Qt::ItemIsEnabled);
//transcript
Transcript transcript = db.longestCodingTranscript(gene_id, Transcript::CCDS);
if (!transcript.isValid()) //fallback to UCSC when no CCDS transcript is defined for the gene
{
transcript = db.longestCodingTranscript(gene_id, Transcript::UCSC);
}
BedFile region = transcript.regions();
setGeneTableItem(r, 1, transcript.name() + " (" + QString::number(region.count()) + " exons)");
//size
region.extend(5);
region.merge();
long long bases = region.baseCount();
setGeneTableItem(r, 2, QString::number(bases), Qt::AlignRight);
//calculate gaps inside target region
BedFile gaps = sys_gaps;
gaps.intersect(region);
//add target region bases not covered by processing system target file
BedFile uncovered(region);
uncovered.subtract(sys_roi);
gaps.add(uncovered);
gaps.merge();
//output (absolute and percentage)
long long gap_bases = gaps.baseCount();
setGeneTableItem(r, 3, QString::number(gap_bases), Qt::AlignRight);
setGeneTableItem(r, 4, QString::number(100.0 * gap_bases / bases, 'f', 2), Qt::AlignRight);
//cnvs + cnv gaps
if (!cna_result_present)
{
setGeneTableItem(r, 5, "n/a", Qt::AlignRight);
setGeneTableItem(r, 6, "n/a", Qt::AlignRight);
}
else
{
BedFile cnv_data = cna_results;
cnv_data.overlapping(region);
cnv_data.sort();
int cnv_del = 0;
int cnv_dup = 0;
int cnv_bad_qc = 0;
for(int i=0; i<cnv_data.count(); ++i)
{
QString cn = cnv_data[i].annotations()[0];
bool ok = false;
int cn_num = cn.toInt(&ok);
if (!ok)
{
++cnv_bad_qc;
}
else if(cn_num<2)
{
++cnv_del;
}
else if(cn_num>2)
{
++cnv_dup;
}
}
QStringList parts;
if (cnv_del) parts << QString::number(cnv_del) + " del";
if (cnv_dup) parts << QString::number(cnv_dup) + " dup";
setGeneTableItem(r, 5, parts.join(", "), Qt::AlignRight, Qt::ItemIsUserCheckable|Qt::ItemIsEnabled);
BedFile region_covered = region;
region_covered.overlapping(cnv_data);
int cnv_gaps = region.count() - region_covered.count();
parts.clear();
if (cnv_bad_qc) parts << QString::number(cnv_bad_qc) + " bad qc";
if (cnv_gaps) parts << QString::number(cnv_gaps) + " not covered";
setGeneTableItem(r, 6, parts.join(", "), Qt::AlignRight);
}
}
//resize
ui->details->resizeColumnsToContents();
ui->details->resizeRowsToContents();
//reset cursor
QApplication::restoreOverrideCursor();
ui->details->blockSignals(false);
updateSelectedGenesStatistics();
}
