private RecalDatum getRecalDatum(
final GATKReportTable reportTable, final int row, final boolean hasEstimatedQReportedColumn) {
final long nObservations =
asLong(reportTable.get(row, RecalUtils.NUMBER_OBSERVATIONS_COLUMN_NAME));
final double nErrors = asDouble(reportTable.get(row, RecalUtils.NUMBER_ERRORS_COLUMN_NAME));
// final double empiricalQuality = asDouble(reportTable.get(row,
// RecalUtils.EMPIRICAL_QUALITY_COLUMN_NAME));
// the estimatedQreported column only exists in the ReadGroup table
final double estimatedQReported =
hasEstimatedQReportedColumn
? (Double) reportTable.get(row, RecalUtils.ESTIMATED_Q_REPORTED_COLUMN_NAME)
: // we get it if we are in the read group table
Byte.parseByte(
(String)
reportTable.get(
row,
RecalUtils
.QUALITY_SCORE_COLUMN_NAME)); // or we use the reported quality if we
// are in any other table
final RecalDatum datum = new RecalDatum(nObservations, nErrors, (byte) 1);
datum.setEstimatedQReported(estimatedQReported);
// datum.setEmpiricalQuality(empiricalQuality); // don't set the value here because we will want
// to recompute with a different conditional Q score prior value
return datum;
}
/**
* Gets the unique read groups in the table
*
* @param report the GATKReport containing the table with RecalUtils.READGROUP_REPORT_TABLE_TITLE
* @return the unique read groups
*/
private static SortedSet<String> getReadGroups(final GATKReport report) {
final GATKReportTable reportTable = report.getTable(RecalUtils.READGROUP_REPORT_TABLE_TITLE);
final SortedSet<String> readGroups = new TreeSet<String>();
for (int i = 0; i < reportTable.getNumRows(); i++)
readGroups.add(reportTable.get(i, RecalUtils.READGROUP_COLUMN_NAME).toString());
return readGroups;
}
/**
* Compiles the list of keys for the ReadGroup table and uses the shared parsing utility to
* produce the actual table
*
* @param reportTable the GATKReport table containing data for this table
* @param rgTable the map representing this table
*/
private void parseReadGroupTable(
final GATKReportTable reportTable, final NestedIntegerArray<RecalDatum> rgTable) {
for (int i = 0; i < reportTable.getNumRows(); i++) {
final Object rg = reportTable.get(i, RecalUtils.READGROUP_COLUMN_NAME);
tempRGarray[0] = requestedCovariates[0].keyFromValue(rg);
final EventType event =
EventType.eventFrom((String) reportTable.get(i, RecalUtils.EVENT_TYPE_COLUMN_NAME));
tempRGarray[1] = event.ordinal();
rgTable.put(getRecalDatum(reportTable, i, true), tempRGarray);
}
}
/**
* Parses the quantization table from the GATK Report and turns it into a map of original =>
* quantized quality scores
*
* @param table the GATKReportTable containing the quantization mappings
* @return an ArrayList with the quantization mappings from 0 to MAX_SAM_QUAL_SCORE
*/
private QuantizationInfo initializeQuantizationTable(GATKReportTable table) {
final Byte[] quals = new Byte[QualityUtils.MAX_SAM_QUAL_SCORE + 1];
final Long[] counts = new Long[QualityUtils.MAX_SAM_QUAL_SCORE + 1];
for (int i = 0; i < table.getNumRows(); i++) {
final byte originalQual = (byte) i;
final Object quantizedObject = table.get(i, RecalUtils.QUANTIZED_VALUE_COLUMN_NAME);
final Object countObject = table.get(i, RecalUtils.QUANTIZED_COUNT_COLUMN_NAME);
final byte quantizedQual = Byte.parseByte(quantizedObject.toString());
final long quantizedCount = Long.parseLong(countObject.toString());
quals[originalQual] = quantizedQual;
counts[originalQual] = quantizedCount;
}
return new QuantizationInfo(Arrays.asList(quals), Arrays.asList(counts));
}
/**
* Compiles the list of keys for the QualityScore table and uses the shared parsing utility to
* produce the actual table
*
* @param reportTable the GATKReport table containing data for this table
* @param qualTable the map representing this table
*/
private void parseQualityScoreTable(
final GATKReportTable reportTable, final NestedIntegerArray<RecalDatum> qualTable) {
for (int i = 0; i < reportTable.getNumRows(); i++) {
final Object rg = reportTable.get(i, RecalUtils.READGROUP_COLUMN_NAME);
tempQUALarray[0] = requestedCovariates[0].keyFromValue(rg);
final Object qual = reportTable.get(i, RecalUtils.QUALITY_SCORE_COLUMN_NAME);
tempQUALarray[1] = requestedCovariates[1].keyFromValue(qual);
final EventType event =
EventType.eventFrom((String) reportTable.get(i, RecalUtils.EVENT_TYPE_COLUMN_NAME));
tempQUALarray[2] = event.ordinal();
qualTable.put(getRecalDatum(reportTable, i, false), tempQUALarray);
}
}
/**
* Parses the arguments table from the GATK Report and creates a RAC object with the proper
* initialization values
*
* @param table the GATKReportTable containing the arguments and its corresponding values
* @return a RAC object properly initialized with all the objects in the table
*/
private RecalibrationArgumentCollection initializeArgumentCollectionTable(GATKReportTable table) {
final RecalibrationArgumentCollection RAC = new RecalibrationArgumentCollection();
for (int i = 0; i < table.getNumRows(); i++) {
final String argument = table.get(i, "Argument").toString();
Object value = table.get(i, RecalUtils.ARGUMENT_VALUE_COLUMN_NAME);
if (value.equals("null"))
value =
null; // generic translation of null values that were printed out as strings | todo --
// add this capability to the GATKReport
if (argument.equals("covariate") && value != null)
RAC.COVARIATES = value.toString().split(",");
else if (argument.equals("standard_covs"))
RAC.DO_NOT_USE_STANDARD_COVARIATES = Boolean.parseBoolean((String) value);
else if (argument.equals("solid_recal_mode"))
RAC.SOLID_RECAL_MODE = RecalUtils.SOLID_RECAL_MODE.recalModeFromString((String) value);
else if (argument.equals("solid_nocall_strategy"))
RAC.SOLID_NOCALL_STRATEGY =
RecalUtils.SOLID_NOCALL_STRATEGY.nocallStrategyFromString((String) value);
else if (argument.equals("mismatches_context_size"))
RAC.MISMATCHES_CONTEXT_SIZE = Integer.parseInt((String) value);
else if (argument.equals("indels_context_size"))
RAC.INDELS_CONTEXT_SIZE = Integer.parseInt((String) value);
else if (argument.equals("mismatches_default_quality"))
RAC.MISMATCHES_DEFAULT_QUALITY = Byte.parseByte((String) value);
else if (argument.equals("insertions_default_quality"))
RAC.INSERTIONS_DEFAULT_QUALITY = Byte.parseByte((String) value);
else if (argument.equals("deletions_default_quality"))
RAC.DELETIONS_DEFAULT_QUALITY = Byte.parseByte((String) value);
else if (argument.equals("maximum_cycle_value"))
RAC.MAXIMUM_CYCLE_VALUE = Integer.parseInt((String) value);
else if (argument.equals("low_quality_tail"))
RAC.LOW_QUAL_TAIL = Byte.parseByte((String) value);
else if (argument.equals("default_platform")) RAC.DEFAULT_PLATFORM = (String) value;
else if (argument.equals("force_platform")) RAC.FORCE_PLATFORM = (String) value;
else if (argument.equals("quantizing_levels"))
RAC.QUANTIZING_LEVELS = Integer.parseInt((String) value);
else if (argument.equals("recalibration_report"))
RAC.existingRecalibrationReport = (value == null) ? null : new File((String) value);
else if (argument.equals("binary_tag_name"))
RAC.BINARY_TAG_NAME = (value == null) ? null : (String) value;
else if (argument.equals("sort_by_all_columns"))
RAC.SORT_BY_ALL_COLUMNS = Boolean.parseBoolean((String) value);
}
return RAC;
}
/**
* Compiles the list of keys for the Covariates table and uses the shared parsing utility to
* produce the actual table
*
* @param reportTable the GATKReport table containing data for this table
* @param recalibrationTables the recalibration tables \
*/
private void parseAllCovariatesTable(
final GATKReportTable reportTable, final RecalibrationTables recalibrationTables) {
for (int i = 0; i < reportTable.getNumRows(); i++) {
final Object rg = reportTable.get(i, RecalUtils.READGROUP_COLUMN_NAME);
tempCOVarray[0] = requestedCovariates[0].keyFromValue(rg);
final Object qual = reportTable.get(i, RecalUtils.QUALITY_SCORE_COLUMN_NAME);
tempCOVarray[1] = requestedCovariates[1].keyFromValue(qual);
final String covName = (String) reportTable.get(i, RecalUtils.COVARIATE_NAME_COLUMN_NAME);
final int covIndex = optionalCovariateIndexes.get(covName);
final Object covValue = reportTable.get(i, RecalUtils.COVARIATE_VALUE_COLUMN_NAME);
tempCOVarray[2] =
requestedCovariates[
RecalibrationTables.TableType.OPTIONAL_COVARIATE_TABLES_START.ordinal() + covIndex]
.keyFromValue(covValue);
final EventType event =
EventType.eventFrom((String) reportTable.get(i, RecalUtils.EVENT_TYPE_COLUMN_NAME));
tempCOVarray[3] = event.ordinal();
recalibrationTables
.getTable(
RecalibrationTables.TableType.OPTIONAL_COVARIATE_TABLES_START.ordinal() + covIndex)
.put(getRecalDatum(reportTable, i, false), tempCOVarray);
}
}