void OrderedDatasetWriter::Open(const string& base_directory, BarcodeDatasets& datasets, int num_regions, const ion::FlowOrder& flow_order, const string& key,
const string& basecaller_name, const string& basecalller_version, const string& basecaller_command_line,
const string& production_date, const string& platform_unit, bool save_filtered_reads)
{
num_regions_ = num_regions;
num_regions_written_ = 0;
region_ready_.assign(num_regions_+1,false);
region_dropbox_.clear();
region_dropbox_.resize(num_regions_);
qv_histogram_.assign(50,0);
num_datasets_ = datasets.num_datasets();
num_barcodes_ = datasets.num_barcodes();
num_read_groups_ = datasets.num_read_groups();
num_reads_.resize(num_datasets_,0);
bam_filename_.resize(num_datasets_);
save_filtered_reads_ = save_filtered_reads;
read_group_name_.resize(num_read_groups_);
read_group_dataset_.assign(num_read_groups_, -1);
read_group_num_Q20_bases_.assign(num_read_groups_,0);
read_group_num_barcode_errors_.resize(num_read_groups_);
for (int rg = 0; rg < num_read_groups_; ++rg) {
read_group_name_[rg] = datasets.read_group_name(rg);
read_group_num_barcode_errors_[rg].assign(3,0);
}
// New filtering and trimming accounting (per read group)
read_group_stats_.resize(num_read_groups_);
bam_writer_.resize(num_datasets_, NULL);
for (int ds = 0; ds < num_datasets_; ++ds) {
// Set up BAM header
bam_filename_[ds] = base_directory + "/" + datasets.dataset(ds)["basecaller_bam"].asString();
SamHeader sam_header;
sam_header.Version = "1.4";
sam_header.SortOrder = "unsorted";
SamProgram sam_program("bc");
sam_program.Name = basecaller_name;
sam_program.Version = basecalller_version;
sam_program.CommandLine = basecaller_command_line;
sam_header.Programs.Add(sam_program);
for (Json::Value::iterator rg = datasets.dataset(ds)["read_groups"].begin(); rg != datasets.dataset(ds)["read_groups"].end(); ++rg) {
string read_group_name = (*rg).asString();
Json::Value& read_group_json = datasets.read_groups()[read_group_name];
read_group_dataset_[datasets.read_group_name_to_id(read_group_name)] = ds;
SamReadGroup read_group (read_group_name);
read_group.FlowOrder = flow_order.full_nucs();
read_group.KeySequence = key;
read_group.KeySequence += read_group_json.get("barcode_sequence","").asString();
read_group.KeySequence += read_group_json.get("barcode_adapter","").asString();
read_group.ProductionDate = production_date;
read_group.Sample = read_group_json.get("sample","").asString();
read_group.Library = read_group_json.get("library","").asString();
read_group.Description = read_group_json.get("description","").asString();
read_group.PlatformUnit = read_group_json.get("platform_unit","").asString();
read_group.SequencingCenter = datasets.json().get("sequencing_center","").asString();
read_group.SequencingTechnology = "IONTORRENT";
sam_header.ReadGroups.Add(read_group);
}
// Open Bam for writing
RefVector empty_reference_vector;
bam_writer_[ds] = new BamWriter();
bam_writer_[ds]->SetCompressionMode(BamWriter::Compressed);
//bam_writer_[ds]->SetCompressionMode(BamWriter::Uncompressed);
bam_writer_[ds]->Open(bam_filename_[ds], sam_header, empty_reference_vector);
}
}
void OrderedDatasetWriter::Open(const string& base_directory, BarcodeDatasets& datasets, int read_class_idx,
int num_regions, const ion::FlowOrder& flow_order, const string& key, const vector<string> & bead_adapters,
int num_bamwriter_threads, const Json::Value & basecaller_json, vector<string>& comments,
MolecularTagTrimmer& tag_trimmer, bool trim_barcodes)
{
num_regions_ = num_regions;
num_regions_written_ = 0;
region_ready_.assign(num_regions_+1,false);
region_dropbox_.clear();
region_dropbox_.resize(num_regions_);
qv_histogram_.assign(50,0);
num_datasets_ = datasets.num_datasets();
num_barcodes_ = datasets.num_barcodes();
num_read_groups_ = datasets.num_read_groups();
num_reads_.resize(num_datasets_,0);
bam_filename_.resize(num_datasets_);
// A negative read group index indicates untrimmed/unfiltered bam files (w. library key) and we save all reads
if (read_class_idx < 0) {
save_filtered_reads_ = true;
read_class_idx = 0;
}
else
save_filtered_reads_ = false;
read_group_name_.resize(num_read_groups_);
read_group_dataset_.assign(num_read_groups_, -1);
read_group_num_Q20_bases_.assign(num_read_groups_,0);
read_group_barcode_filt_zero_err_.assign(num_read_groups_, 0);
read_group_barcode_adapter_rejected_.assign(num_read_groups_, 0);
read_group_num_barcode_errors_.resize(num_read_groups_);
read_group_barcode_distance_hist_.resize(num_read_groups_);
read_group_barcode_bias_.resize(num_read_groups_);
for (int rg = 0; rg < num_read_groups_; ++rg) {
read_group_name_[rg] = datasets.read_group_name(rg);
read_group_num_barcode_errors_[rg].assign(3,0);
read_group_barcode_bias_[rg].assign(datasets.GetBCmaxFlows(),0.0);
read_group_barcode_distance_hist_[rg].assign(5,0);
}
// New filtering and trimming accounting (per read group)
read_group_stats_.resize(num_read_groups_);
for (int rg=0; rg<num_read_groups_; rg++)
read_group_stats_[rg].SetBeadAdapters(bead_adapters);
combined_stats_.SetBeadAdapters(bead_adapters);
bam_writer_.resize(num_datasets_, NULL);
sam_header_.resize(num_datasets_);
num_bamwriter_threads_ = num_bamwriter_threads;
for (int ds = 0; ds < num_datasets_; ++ds) {
// Set up BAM header
bam_filename_[ds] = base_directory + "/" + datasets.dataset(ds)["basecaller_bam"].asString();
SamHeader& sam_header = sam_header_[ds];
sam_header.Version = "1.4";
sam_header.SortOrder = "unsorted";
SamProgram sam_program("bc");
sam_program.Name = "BaseCaller";
sam_program.Version = basecaller_json["BaseCaller"]["version"].asString() + "/" + basecaller_json["BaseCaller"]["git_hash"].asString();
sam_program.CommandLine = basecaller_json["BaseCaller"]["command_line"].asString();
sam_header.Programs.Add(sam_program);
for (Json::Value::iterator rg = datasets.dataset(ds)["read_groups"].begin(); rg != datasets.dataset(ds)["read_groups"].end(); ++rg) {
string read_group_name = (*rg).asString();
Json::Value& read_group_json = datasets.read_groups()[read_group_name];
read_group_dataset_[datasets.read_group_name_to_id(read_group_name)] = ds;
SamReadGroup read_group (read_group_name);
read_group.FlowOrder = flow_order.full_nucs();
read_group.KeySequence = key;
if (trim_barcodes){ // We only add the barcode info to the key sequence if we hard clipped it
read_group.KeySequence += read_group_json.get("barcode_sequence","").asString();
read_group.KeySequence += read_group_json.get("barcode_adapter","").asString();
}
read_group.ProductionDate = basecaller_json["BaseCaller"]["start_time"].asString();
read_group.Sample = read_group_json.get("sample","").asString();
read_group.Library = read_group_json.get("library","").asString();
read_group.Description = read_group_json.get("description","").asString();
read_group.PlatformUnit = read_group_json.get("platform_unit","").asString();
read_group.SequencingCenter = datasets.json().get("sequencing_center","").asString();
read_group.SequencingTechnology = "IONTORRENT";
// Add custom tags: Structure of tags per read group XXX
if (datasets.IsLibraryDataset()) {
MolTag my_tags = tag_trimmer.GetReadGroupTags(read_group_name);
AddCustomReadGroupTag(read_group, "zt", my_tags.prefix_mol_tag);
AddCustomReadGroupTag(read_group, "yt", my_tags.suffix_mol_tag);
}
sam_header.ReadGroups.Add(read_group);
}
for(size_t i = 0; i < comments.size(); ++i)
sam_header.Comments.push_back(comments[i]);
}
}