TEST(BamHeaderTest, ExtractFromRawDataOk)
{
ReadGroupInfo rg1("rg1");
rg1.Sample("control");
ReadGroupInfo rg2("rg2");
rg2.Sample("condition1");
ReadGroupInfo rg3("rg3");
rg3.Sample("condition1");
SequenceInfo seq1("chr1");
seq1.Length("2038").Species("chocobo");
SequenceInfo seq2("chr2");
seq2.Length("3042").Species("chocobo");
ProgramInfo prog1("_foo_");
prog1.Name("ide");
BamHeader header;
header.Version("1.1")
.SortOrder("queryname")
.PacBioBamVersion("3.0.1")
.AddReadGroup(rg1)
.AddReadGroup(rg2)
.AddReadGroup(rg3)
.AddSequence(seq1)
.AddSequence(seq2)
.AddProgram(prog1)
.AddComment("ipsum and so on")
.AddComment("citation needed");
const string& expectedText = "@HD\tVN:1.1\tSO:queryname\tpb:3.0.1\n"
"@SQ\tSN:chr1\tLN:2038\tSP:chocobo\n"
"@SQ\tSN:chr2\tLN:3042\tSP:chocobo\n"
"@RG\tID:rg1\tPL:PACBIO\tDS:READTYPE=UNKNOWN\tSM:control\tPM:SEQUEL\n"
"@RG\tID:rg2\tPL:PACBIO\tDS:READTYPE=UNKNOWN\tSM:condition1\tPM:SEQUEL\n"
"@RG\tID:rg3\tPL:PACBIO\tDS:READTYPE=UNKNOWN\tSM:condition1\tPM:SEQUEL\n"
"@PG\tID:_foo_\tPN:ide\n"
"@CO\tipsum and so on\n"
"@CO\tcitation needed\n";
string text = header.ToSam();
PBBAM_SHARED_PTR<bam_hdr_t> rawData(sam_hdr_parse(text.size(), text.c_str()), tests::BamHdrDeleter());
rawData->ignore_sam_err = 0;
rawData->cigar_tab = NULL;
rawData->l_text = text.size();
rawData->text = (char*)calloc(rawData->l_text + 1, 1);
memcpy(rawData->text, text.c_str(), rawData->l_text);
const BamHeader newHeader = BamHeader(string(rawData->text, rawData->l_text));
EXPECT_EQ(header.Version(), newHeader.Version());
EXPECT_EQ(header.SortOrder(), newHeader.SortOrder());
EXPECT_EQ(header.PacBioBamVersion(), newHeader.PacBioBamVersion());
text = newHeader.ToSam();
EXPECT_EQ(expectedText, text);
}
BamHeader PrepareHeader(const OptionParser& parser, int argc, char** argv,
const vector<string>& files)
{
using boost::algorithm::join;
ProgramInfo program(parser.prog() + "-" + CCS_VERSION);
program.Name(parser.prog())
.CommandLine(parser.prog() + " " + join(vector<string>(argv + 1, argv + argc), " "))
.Description(DESCRIPTION)
.Version(CCS_VERSION);
BamHeader header;
header.PacBioBamVersion("3.0.1").SortOrder("unknown").Version("1.5").AddProgram(program);
for (const auto& file : files) {
BamFile bam(file);
for (const auto& rg : bam.Header().ReadGroups()) {
if (rg.ReadType() != "SUBREAD")
parser.error("invalid input file, READTYPE must be SUBREAD");
ReadGroupInfo readGroup(rg.MovieName(), "CCS");
readGroup.BindingKit(rg.BindingKit())
.SequencingKit(rg.SequencingKit())
.BasecallerVersion(rg.BasecallerVersion())
.FrameRateHz(rg.FrameRateHz());
header.AddReadGroup(readGroup);
}
}
return header;
}
TEST(BamHeaderTest, DecodeTest)
{
const string& text = "@HD\tVN:1.1\tSO:queryname\tpb:3.0.1\n"
"@SQ\tSN:chr1\tLN:2038\tSP:chocobo\n"
"@SQ\tSN:chr2\tLN:3042\tSP:chocobo\n"
"@RG\tID:rg1\tSM:control\n"
"@RG\tID:rg2\tSM:condition1\n"
"@RG\tID:rg3\tSM:condition1\n"
"@PG\tID:_foo_\tPN:ide\n"
"@CO\tipsum and so on\n"
"@CO\tcitation needed\n";
BamHeader header = BamHeader(text);
EXPECT_EQ(string("1.1"), header.Version());
EXPECT_EQ(string("queryname"), header.SortOrder());
EXPECT_EQ(string("3.0.1"), header.PacBioBamVersion());
EXPECT_EQ(3, header.ReadGroups().size());
EXPECT_TRUE(header.HasReadGroup("rg1"));
EXPECT_TRUE(header.HasReadGroup("rg2"));
EXPECT_TRUE(header.HasReadGroup("rg3"));
EXPECT_EQ(string("control"), header.ReadGroup("rg1").Sample());
EXPECT_EQ(string("condition1"), header.ReadGroup("rg2").Sample());
EXPECT_EQ(string("condition1"), header.ReadGroup("rg3").Sample());
EXPECT_EQ(2, header.Sequences().size());
EXPECT_TRUE(header.HasSequence("chr1"));
EXPECT_TRUE(header.HasSequence("chr2"));
EXPECT_EQ(string("chocobo"), header.Sequence("chr1").Species());
EXPECT_EQ(string("chocobo"), header.Sequence("chr2").Species());
EXPECT_EQ(string("2038"), header.Sequence("chr1").Length());
EXPECT_EQ(string("3042"), header.Sequence("chr2").Length());
EXPECT_EQ(1, header.Programs().size());
EXPECT_TRUE(header.HasProgram("_foo_"));
EXPECT_EQ(string("ide"), header.Program("_foo_").Name());
EXPECT_EQ(2, header.Comments().size());
EXPECT_EQ(string("ipsum and so on"), header.Comments().at(0));
EXPECT_EQ(string("citation needed"), header.Comments().at(1));
}
static BamHeader PrepareHeader(const std::string& cmdLine,
const std::vector<ReadGroupInfo>& inputReadgroups)
{
ProgramInfo program{APPNAME + "-" + PacBio::UnanimityVersion()};
program.Name(APPNAME)
.CommandLine(APPNAME + " " + cmdLine)
.Description(DESCRIPTION)
.Version(PacBio::UnanimityVersion());
std::random_device rd;
std::default_random_engine rng{rd()};
if (inputReadgroups.size() > 1) {
std::cerr << APPNAME << " can currently only handle one readgroup per input bam\n";
exit(EXIT_FAILURE);
}
std::ostringstream movieName;
movieName << 'm' << std::uniform_int_distribution<uint32_t>{10000, 99999}(rng) << '_'
<< std::uniform_int_distribution<uint32_t>{100000, 999999}(rng) << '_'
<< std::uniform_int_distribution<uint32_t>{100000, 999999}(rng);
// reuse input readgroups for platform information
ReadGroupInfo newRg{inputReadgroups.front()};
newRg.MovieName(movieName.str())
.ReadType("SUBREAD")
.PlatformModel(PlatformModelType::SEQUEL)
.IpdCodec(FrameCodec::V1)
.PulseWidthCodec(FrameCodec::V1)
.Id(movieName.str(), "SUBREAD");
BamHeader header;
header.PacBioBamVersion("3.0.1")
.SortOrder("unknown")
.Version("1.5")
.AddProgram(program)
.ReadGroups(std::vector<ReadGroupInfo>{newRg});
return header;
}
