TEST(BamHeaderTest, DefaultConstruction)
{
BamHeader header;
EXPECT_TRUE(header.Version().empty());
EXPECT_TRUE(header.SortOrder().empty()); // default to unknown ?
EXPECT_TRUE(header.ReadGroups().empty());
EXPECT_TRUE(header.Sequences().empty());
EXPECT_TRUE(header.Programs().empty());
EXPECT_TRUE(header.Comments().empty());
EXPECT_THROW(header.Program("foo"), std::exception);
EXPECT_THROW(header.ReadGroup("foo"), std::exception);
EXPECT_THROW(header.SequenceId("foo"), std::exception);
EXPECT_THROW(header.SequenceLength(42), std::exception);
EXPECT_THROW(header.SequenceName(42), std::exception);
}
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));
}
TEST(PacBioIndexTest, CreateOnTheFly)
{
// do this in temp directory, so we can ensure write access
const string tempDir = tests::GeneratedData_Dir + "/";
const string tempBamFn = tempDir + "temp.bam";
const string tempPbiFn = tempBamFn + ".pbi";
// NOTE: new file differs in size than existing (different write parameters may yield different file sizes, even though content is same)
const vector<int64_t> expectedNewOffsets = { 33816576, 236126208, 391315456, 469106688, 537067520, 587792384, 867303424, 1182793728, 1449787392, 1582628864 };
vector<int64_t> observedOffsets;
// create PBI on the fly from input BAM while we write to new file
{
BamFile bamFile(test2BamFn);
BamHeader header = bamFile.Header();
BamWriter writer(tempBamFn, header); // default compression, default thread count
PbiBuilder builder(tempPbiFn, header.Sequences().size());
int64_t vOffset = 0;
EntireFileQuery entireFile(bamFile);
for (const BamRecord& record : entireFile) {
writer.Write(record, &vOffset);
builder.AddRecord(record, vOffset);
observedOffsets.push_back(vOffset);
}
}
EXPECT_EQ(expectedNewOffsets, observedOffsets);
// sanity check on original file
{
const vector<int64_t> originalFileOffsets = { 33816576, 33825163, 33831333, 33834264, 33836542, 33838065, 33849818, 33863499, 33874621, 1392836608 };
BamRecord r;
BamReader reader(test2BamFn);
for (int i = 0; i < originalFileOffsets.size(); ++i) {
reader.VirtualSeek(originalFileOffsets.at(i));
EXPECT_TRUE(CanRead(reader, r, i));
}
}
// attempt to seek in our new file using both expected & observed offsets
{
BamRecord r;
BamReader reader(tempBamFn);
for (int i = 0; i < expectedNewOffsets.size(); ++i) {
reader.VirtualSeek(expectedNewOffsets.at(i));
EXPECT_TRUE(CanRead(reader, r, i));
}
for (int i = 0; i < observedOffsets.size(); ++i) {
reader.VirtualSeek(observedOffsets.at(i));
EXPECT_TRUE(CanRead(reader, r, i));
}
}
// compare data in new PBI file, to expected data
const PbiRawData& expectedIndex = tests::Test2Bam_NewIndex();
const PbiRawData& fromBuilt = PbiRawData(tempPbiFn);
tests::ExpectRawIndicesEqual(expectedIndex, fromBuilt);
// straight diff of newly-generated PBI file to existing PBI
// TODO: Come back to this once pbindexump is in place.
// We can't exactly do this since file offsets may differ between 2 BAMs of differing compression levels.
// Should add some sort of BAM checksum based on contents, not just size, for this reason.
// const string pbiDiffCmd = string("diff -q ") + test2BamFn + ".pbi " + tempPbiFn;
// EXPECT_EQ(0, system(pbiDiffCmd.c_str()));
// clean up temp file(s)
remove(tempBamFn.c_str());
remove(tempPbiFn.c_str());
}
