/// <summary>
/// load the format/info meta data maps (these are used for quick lookup by key name)
/// </summary>
private void loadMetaDataMaps()
{
foreach (VCFHeaderLine line in mMetaData)
{
if (line is VCFInfoHeaderLine)
{
VCFInfoHeaderLine infoLine = (VCFInfoHeaderLine)line;
addMetaDataMapBinding(mInfoMetaData, infoLine);
}
else if (line is VCFFormatHeaderLine)
{
VCFFormatHeaderLine formatLine = (VCFFormatHeaderLine)line;
addMetaDataMapBinding(mFormatMetaData, formatLine);
}
else if (line is VCFFilterHeaderLine)
{
VCFFilterHeaderLine filterLine = (VCFFilterHeaderLine)line;
mFilterMetaData[filterLine.ID] = filterLine;
}
else if (line is VCFContigHeaderLine)
{
contigMetaData.Add((VCFContigHeaderLine)line);
}
else
{
mOtherMetaData[line.Key] = line;
}
}
if (hasFormatLine(VCFConstants.GENOTYPE_LIKELIHOODS_KEY) && !hasFormatLine(VCFConstants.GENOTYPE_PL_KEY))
{
Console.WriteLine("Warning now we want PL fields, not just GL fields");
//throw new VCFParsingError("Found " + VCFConstants.GENOTYPE_LIKELIHOODS_KEY + " format, but no " + VCFConstants.GENOTYPE_PL_KEY + " field. We now only manage PL fields internally.");
}
}
/// <summary>
/// add the genotype data
/// </summary>
/// <param name="vc"> the variant context </param>
/// <param name="genotypeFormatKeys"> Genotype formatting string </param>
/// <param name="alleleMap"> alleles for this context </param>
/// <exception cref="IOException"> for writer </exception>
private string getGenotypeDataText(VariantContext vc, IDictionary <Allele, string> alleleMap, IList <string> genotypeFormatKeys)
{
StringBuilder sbn = new StringBuilder();
int ploidy = vc.GetMaxPloidy(2);
foreach (string sample in mHeader.GenotypeSampleNames)
{
sbn.Append(VCFConstants.FIELD_SEPARATOR);
Genotype g = vc.GetGenotype(sample);
if (g == null)
{
g = GenotypeBuilder.CreateMissing(sample, ploidy);
}
IList <string> attrs = new List <string>(genotypeFormatKeys.Count);
foreach (string field in genotypeFormatKeys)
{
if (field.Equals(VCFConstants.GENOTYPE_KEY))
{
if (!g.Available)
{
throw new Exception("GTs cannot be missing for some samples if they are available for others in the record");
}
sbn.Append(getAlleleText(g.getAllele(0), alleleMap));
for (int i = 1; i < g.Ploidy; i++)
{
sbn.Append(g.Phased ? VCFConstants.PHASED : VCFConstants.UNPHASED);
sbn.Append(getAlleleText(g.getAllele(i), alleleMap));
}
continue;
}
else
{
string outputValue;
if (field.Equals(VCFConstants.GENOTYPE_FILTER_KEY))
{
outputValue = g.Filtered ? g.Filters : VCFConstants.PASSES_FILTERS_v4;
}
else
{
IntGenotypeFieldAccessors.Accessor accessor = intGenotypeFieldAccessors.GetAccessor(field);
if (accessor != null)
{
int[] intValues = accessor.getValues(g);
if (intValues == null)
{
outputValue = VCFConstants.MISSING_VALUE_v4;
}
else if (intValues.Length == 1) // fast path
{
outputValue = Convert.ToString(intValues[0]);
}
else
{
StringBuilder sb = new StringBuilder();
sb.Append(intValues[0]);
for (int i = 1; i < intValues.Length; i++)
{
sb.Append(",");
sb.Append(intValues[i]);
}
outputValue = sb.ToString();
}
}
else
{
object val = g.HasExtendedAttribute(field) ? g.GetExtendedAttribute(field) : VCFConstants.MISSING_VALUE_v4;
VCFFormatHeaderLine metaData = mHeader.getFormatHeaderLine(field);
if (metaData != null)
{
int numInFormatField = metaData.getCount(vc);
if (numInFormatField > 1 && val.Equals(VCFConstants.MISSING_VALUE_v4))
{
// If we have a missing field but multiple values are expected, we need to construct a new string with all fields.
// For example, if Number=2, the string has to be ".,."
StringBuilder sb = new StringBuilder(VCFConstants.MISSING_VALUE_v4);
for (int i = 1; i < numInFormatField; i++)
{
sb.Append(",");
sb.Append(VCFConstants.MISSING_VALUE_v4);
}
val = sb.ToString();
}
}
// assume that if key is absent, then the given string encoding suffices
outputValue = formatVCFField(val);
}
}
if (outputValue != null)
{
attrs.Add(outputValue);
}
}
}
// strip off trailing missing values
for (int i = attrs.Count - 1; i >= 0; i--)
{
if (isMissingValue(attrs[i]))
{
attrs.RemoveAt(i);
}
else
{
break;
}
}
for (int i = 0; i < attrs.Count; i++)
{
if (i > 0 || genotypeFormatKeys.Contains(VCFConstants.GENOTYPE_KEY))
{
sbn.Append(VCFConstants.GENOTYPE_FIELD_SEPARATOR);
}
sbn.Append(attrs[i]);
}
}
return(sbn.ToString());
}
/// <summary>
/// create a VCF header from a set of header record lines
/// </summary>
/// <param name="headerStrings"> a list of strings that represent all the ## and # entries </param>
/// <returns> a VCFHeader object </returns>
protected internal virtual VCFHeader parseHeaderFromLines (IList<string> headerStrings, VCFHeaderVersion version)
{
this.version = version;
ISet<VCFHeaderLine> metaData = new LinkedHashSet<VCFHeaderLine> ();
ISet<string> sampleNames = new LinkedHashSet<string> ();
int contigCounter = 0;
// iterate over all the passed in strings
foreach (string str in headerStrings) {
if (!str.StartsWith (VCFHeader.METADATA_INDICATOR)) {//presumably the #CHROM POS ID REF ALT QUAL FILTER INFO etc. line
string[] strings = str.Substring (1).Split (VCFConstants.FIELD_SEPARATOR_CHAR);
//check for null last string, grrr...
if (String.IsNullOrEmpty (strings.Last ())) {
strings = strings.Take (strings.Length - 1).ToArray ();
}
if (strings.Length < VCFHeader.HEADER_FIELDS.Length) {
throw new VCFParsingError ("There are not enough columns present in the header line: " + str);
}
//Verify Arrays
var misMatchedColumns = Enumerable.Range (0, VCFHeader.HEADER_FIELDS.Length).Where (x => VCFHeader.HEADER_FIELDS [x] != strings [x]).Select (x => strings [x]).ToArray ();
if (misMatchedColumns.Length > 0) {
throw new VCFParsingError ("We were not expecting column name '" + misMatchedColumns [0] + " in that position");
}
int arrayIndex = VCFHeader.HEADER_FIELDS.Length;//start after verified columns
bool sawFormatTag = false;
if (arrayIndex < strings.Length) {
if (!strings [arrayIndex].Equals ("FORMAT")) {
throw new VCFParsingError ("we were expecting column name 'FORMAT' but we saw '" + strings [arrayIndex] + "'");
}
sawFormatTag = true;
arrayIndex++;
}
while (arrayIndex < strings.Length) {
sampleNames.Add (strings [arrayIndex++]);
}
if (sawFormatTag && sampleNames.Count == 0) {
throw new VCFParsingError ("The FORMAT field was provided but there is no genotype/sample data");
}
} else {
if (str.StartsWith (VCFConstants.INFO_HEADER_START)) {
VCFInfoHeaderLine info = new VCFInfoHeaderLine (str.Substring (7), version);
metaData.Add (info);
} else if (str.StartsWith (VCFConstants.FILTER_HEADER_START)) {
VCFFilterHeaderLine filter = new VCFFilterHeaderLine (str.Substring (9), version);
metaData.Add (filter);
} else if (str.StartsWith (VCFConstants.FORMAT_HEADER_START)) {
VCFFormatHeaderLine format = new VCFFormatHeaderLine (str.Substring (9), version);
metaData.Add (format);
} else if (str.StartsWith (VCFConstants.CONTIG_HEADER_START)) {
VCFContigHeaderLine contig = new VCFContigHeaderLine (str.Substring (9), version, VCFConstants.CONTIG_HEADER_START.Substring (2), contigCounter++);
metaData.Add (contig);
} else if (str.StartsWith (VCFConstants.ALT_HEADER_START)) {
//TODO: Consider giving Alt header lines their own class
VCFSimpleHeaderLine alt = new VCFSimpleHeaderLine (str.Substring (6), version, VCFConstants.ALT_HEADER_START.Substring (2), "ID", "Description");
metaData.Add (alt);
} else {
int equals = str.IndexOf ("=");
if (equals != -1) {
metaData.Add (new VCFHeaderLine (str.Substring (2, equals - 2), str.Substring (equals + 1)));
}
}
}
}
this.header = new VCFHeader (metaData, sampleNames);
if (doOnTheFlyModifications) {
this.header = VCFStandardHeaderLines.repairStandardHeaderLines (this.header);
}
return this.header;
}
/// <summary>
/// create a VCF header from a set of header record lines
/// </summary>
/// <param name="headerStrings"> a list of strings that represent all the ## and # entries </param>
/// <returns> a VCFHeader object </returns>
protected internal virtual VCFHeader parseHeaderFromLines(IList <string> headerStrings, VCFHeaderVersion version)
{
this.version = version;
ISet <VCFHeaderLine> metaData = new LinkedHashSet <VCFHeaderLine> ();
ISet <string> sampleNames = new LinkedHashSet <string> ();
int contigCounter = 0;
// iterate over all the passed in strings
foreach (string str in headerStrings)
{
if (!str.StartsWith(VCFHeader.METADATA_INDICATOR)) //presumably the #CHROM POS ID REF ALT QUAL FILTER INFO etc. line
{
string[] strings = str.Substring(1).Split(VCFConstants.FIELD_SEPARATOR_CHAR);
//check for null last string, grrr...
if (String.IsNullOrEmpty(strings.Last()))
{
strings = strings.Take(strings.Length - 1).ToArray();
}
if (strings.Length < VCFHeader.HEADER_FIELDS.Length)
{
throw new VCFParsingError("There are not enough columns present in the header line: " + str);
}
//Verify Arrays
var misMatchedColumns = Enumerable.Range(0, VCFHeader.HEADER_FIELDS.Length).Where(x => VCFHeader.HEADER_FIELDS [x] != strings [x]).Select(x => strings [x]).ToArray();
if (misMatchedColumns.Length > 0)
{
throw new VCFParsingError("We were not expecting column name '" + misMatchedColumns [0] + " in that position");
}
int arrayIndex = VCFHeader.HEADER_FIELDS.Length; //start after verified columns
bool sawFormatTag = false;
if (arrayIndex < strings.Length)
{
if (!strings [arrayIndex].Equals("FORMAT"))
{
throw new VCFParsingError("we were expecting column name 'FORMAT' but we saw '" + strings [arrayIndex] + "'");
}
sawFormatTag = true;
arrayIndex++;
}
while (arrayIndex < strings.Length)
{
sampleNames.Add(strings [arrayIndex++]);
}
if (sawFormatTag && sampleNames.Count == 0)
{
throw new VCFParsingError("The FORMAT field was provided but there is no genotype/sample data");
}
}
else
{
if (str.StartsWith(VCFConstants.INFO_HEADER_START))
{
VCFInfoHeaderLine info = new VCFInfoHeaderLine(str.Substring(7), version);
metaData.Add(info);
}
else if (str.StartsWith(VCFConstants.FILTER_HEADER_START))
{
VCFFilterHeaderLine filter = new VCFFilterHeaderLine(str.Substring(9), version);
metaData.Add(filter);
}
else if (str.StartsWith(VCFConstants.FORMAT_HEADER_START))
{
VCFFormatHeaderLine format = new VCFFormatHeaderLine(str.Substring(9), version);
metaData.Add(format);
}
else if (str.StartsWith(VCFConstants.CONTIG_HEADER_START))
{
VCFContigHeaderLine contig = new VCFContigHeaderLine(str.Substring(9), version, VCFConstants.CONTIG_HEADER_START.Substring(2), contigCounter++);
metaData.Add(contig);
}
else if (str.StartsWith(VCFConstants.ALT_HEADER_START))
{
//TODO: Consider giving Alt header lines their own class
VCFSimpleHeaderLine alt = new VCFSimpleHeaderLine(str.Substring(6), version, VCFConstants.ALT_HEADER_START.Substring(2), "ID", "Description");
metaData.Add(alt);
}
else
{
int equals = str.IndexOf("=");
if (equals != -1)
{
metaData.Add(new VCFHeaderLine(str.Substring(2, equals - 2), str.Substring(equals + 1)));
}
}
}
}
this.header = new VCFHeader(metaData, sampleNames);
if (doOnTheFlyModifications)
{
this.header = VCFStandardHeaderLines.repairStandardHeaderLines(this.header);
}
return(this.header);
}
private static void registerStandard(VCFFormatHeaderLine line)
{
formatStandards.add(line);
}
