TEST(Fastq, DefaultConstructor)
{
Fastq fq;
EXPECT_EQ(0, fq.name().compare(""));
EXPECT_EQ(0, fq.seq().compare(""));
EXPECT_EQ(0, fq.qual().compare(""));
}
TEST(Fastq, ReadFromFile)
{
Fastq fq;
unsigned int counter = 0;
ifstream inStream("test_files/fastq_unittest.fastq");
if (! inStream.is_open())
{
cerr << "Error opening test file test_files/fastq_unittest.fastq" << endl;
exit(1);
}
while (fq.fillFromFile(inStream))
{
counter++;
string expectedName = static_cast<ostringstream*>( &(ostringstream() << counter) )->str();
EXPECT_EQ(0, fq.name().compare(expectedName));
EXPECT_EQ(0, fq.seq().compare("ACGT"));
}
}
TEST(Fastq, setValues)
{
Fastq fq;
fq.name("name");
fq.seq("ACGT");
fq.qual("IIII");
EXPECT_EQ(0, fq.name().compare("name"));
EXPECT_EQ(0, fq.seq().compare("ACGT"));
EXPECT_EQ(0, fq.qual().compare("IIII"));
}
void check_fastq_fields(const Fastq& record, const string& name, const string& comment, const string& sequence, const string& quals) {
BOOST_CHECK_EQUAL(record.name() , name);
BOOST_CHECK_EQUAL(record.comment() , comment);
BOOST_CHECK_EQUAL(record.sequence(), sequence);
BOOST_CHECK_EQUAL(record.quals() , quals);
}
int main( int argc, char *argv[])
{
// Options
string readfile;
string prefix = "result";
int winsize = 6;
int minlen = 30;
int threshold = 30;
int offset = PHREDOFFSET;
bool showHelp = false;
// Show help when has no options
if(argc <= 1)
{
Help();
return 0;
}
// Parsing options
for(int i = 1; i < argc; i++)
{
int parameterLength = (int)strlen(argv[i]);
if((PARAMETER_CHECK("-h", 2, parameterLength)) || (PARAMETER_CHECK("--help", 5, parameterLength)))
showHelp=true;
else if((PARAMETER_CHECK("-i", 2, parameterLength)) || (PARAMETER_CHECK("--input", 7, parameterLength)))
{
if ((++i) < argc)
readfile = argv[i];
}
else if((PARAMETER_CHECK("-w", 2, parameterLength)) || (PARAMETER_CHECK("--winsize", 9, parameterLength)))
{
if ((++i) < argc)
winsize = StringUtils::toValue<int>(argv[i]);
}
else if ((PARAMETER_CHECK("-m", 2, parameterLength)) || (PARAMETER_CHECK("--minlen", 8, parameterLength)))
{
if ((++i) < argc)
minlen = StringUtils::toValue<int>(argv[i]);
}
else if ((PARAMETER_CHECK("-p", 2, parameterLength)) || (PARAMETER_CHECK("--prefix", 8, parameterLength)))
{
if ((++i) < argc)
prefix = argv[i];
}
else if ((PARAMETER_CHECK("-o", 2, parameterLength)) || (PARAMETER_CHECK("--offset", 8, parameterLength)))
{
if ((++i) < argc)
offset = StringUtils::toValue<int>(argv[i]);
}
else if ((PARAMETER_CHECK("-t", 2, parameterLength)) || (PARAMETER_CHECK("--threshold", 11, parameterLength)))
{
if ((++i) < argc)
{
threshold = StringUtils::toValue<int>(argv[i]);
if (threshold <0 || threshold >50)
{
cerr << endl << "*****ERROR: Threshold should be within 0~50, reset to default: 30. *****" << endl << endl;
threshold = 30;
}
}
}
else
{
cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
showHelp = true;
}
}
// Show help if no proper auguments.
if (showHelp)
{
Help();
return 0;
}
// Variables
int counter=0,lowcounter=0,i;
int qsum=0,sumThreshold;
char buf[MAXSEQREADLEN]; // Store the sequence for trimming
int start,seqlen;
Fastq seq;
SeqReader fq(readfile);
Writer trimmed(prefix+"_trimmed.fq");
Writer lowqual(prefix+"_low_qual.fq");
// open files
fq.open();
trimmed.open();
lowqual.open();
// Read the fastq file.
sumThreshold=(threshold+offset)*winsize;
while (fq.getNext(seq))
{
if (++counter%100000==0)
cerr << "\rFinished " << counter << " reads.";
strcpy(buf,seq.phred.c_str());
seqlen=seq.length();
qsum=0;
start=0;
// Trim 5' ends
for(i=0;i<winsize;i++)
qsum+=buf[i];
while(qsum<sumThreshold)
{
start++;
if (seqlen-start< minlen)
break;
qsum+=buf[start+winsize-1]-buf[start-1];
}
// Trim 3' ends
if (seqlen-start >= minlen)
{
qsum=0;
for( i=seqlen-1;i>=seqlen-winsize;i--)
qsum+=buf[i];
while (qsum<sumThreshold)
{
seqlen--; // trim the last nucleotide.
if (seqlen-start < minlen)
break;
qsum+=buf[seqlen-winsize-1]-buf[seqlen-1];
}
}
if (seqlen-start>=minlen)
{
seq.seq=seq.seq.substr(start,seqlen-start);
seq.phred=seq.phred.substr(start,seqlen-start);
(*(trimmed.Printer())) << seq;
}
else
{
(*(lowqual.Printer())) << seq;
lowcounter++;
}
}
// close files
fq.close();
trimmed.close();
lowqual.close();
// Print summary
cerr << endl;
cerr << "Parameters: \n\tthreshold = " << threshold << "\n\twindow size = " << winsize << "\n\tminimum length = " << minlen << endl;
cerr << "Summary:" << endl;
cerr << "\tTotal reads : " << counter << endl;
cerr << "\tTrimmed reads: " << counter-lowcounter << endl;
cerr << "\tShort reads : " << lowcounter <<endl;
return 0;
}
