/// <summary>
/// Create the info string; assumes that no values are null
/// </summary>
/// <param name="infoFields"> a map of info fields </param>
/// <exception cref="IOException"> for writer </exception>
private string getInfoString(IDictionary <string, string> infoFields)
{
if (infoFields.Count == 0)
{
return(VCFConstants.EMPTY_INFO_FIELD);
}
bool isFirst = true;
string toReturn = "";
foreach (KeyValuePair <string, string> entry in infoFields)
{
if (isFirst)
{
isFirst = false;
}
else
{
toReturn += VCFConstants.INFO_FIELD_SEPARATOR;
}
string key = entry.Key;
toReturn += key;
if (!entry.Value.Equals(""))
{
VCFInfoHeaderLine metaData = mHeader.getInfoHeaderLine(key);
if (metaData == null || metaData.CountType != VCFHeaderLineCount.INTEGER || metaData.Count != 0)
{
toReturn += "=";
toReturn += entry.Value;
}
}
}
return(toReturn);
}
/// <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>
/// parse out the info fields into a dictionary of key/values </summary>
/// <param name="infoField"> the fields </param>
/// <returns> a mapping of keys to objects </returns>
private IDictionary <string, object> parseInfo(string infoField)
{
IDictionary <string, object> attributes = new Dictionary <string, object> ();
if (infoField.Length == 0)
{
generateException("The VCF specification requires a valid info field");
}
if (!infoField.Equals(VCFConstants.EMPTY_INFO_FIELD))
{
if (infoField.IndexOf("\t") != -1 || infoField.IndexOf(" ") != -1)
{
generateException("The VCF specification does not allow for whitespace in the INFO field");
}
string[] infoFieldArray = infoField.Split(VCFConstants.INFO_FIELD_SEPARATOR_CHAR_AS_ARRAY);
//int infoFieldSplitSize = ParsingUtils.Split(infoField, infoFieldArray, VCFConstants.INFO_FIELD_SEPARATOR_CHAR, false);
for (int i = 0; i < infoFieldArray.Length; i++)
{
string key;
object value;
int eqI = infoFieldArray [i].IndexOf("=");
if (eqI != -1)
{
key = infoFieldArray [i].Substring(0, eqI);
string valueString = infoFieldArray [i].Substring(eqI + 1);
// split on the INFO field separator
string[] infoValueArray = valueString.Split(VCFConstants.INFO_FIELD_ARRAY_SEPARATOR_CHAR);
//int infoValueSplitSize = ParsingUtils.Split(valueString, infoValueArray, VCFConstants.INFO_FIELD_ARRAY_SEPARATOR_CHAR, false);
if (infoValueArray.Length == 1) // infoValueSplitSize == 1)
{
value = infoValueArray [0];
VCFInfoHeaderLine headerLine = header.getInfoHeaderLine(key);
if (headerLine != null && headerLine.Type == VCFHeaderLineType.Flag && value.Equals("0"))
{
// deal with the case where a flag field has =0, such as DB=0, by skipping the add
continue;
}
}
else
{
//List<string> valueList = new List<string>(infoValueArray);
value = infoValueArray;
}
}
else
{
key = infoFieldArray [i];
VCFInfoHeaderLine headerLine = header.getInfoHeaderLine(key);
if (headerLine != null && headerLine.Type != VCFHeaderLineType.Flag)
{
//Note: This was originally a warning that was only thrown once
throw new VCFParsingError("Found info key " + key + " without a = value, but the header says the field is of type " + headerLine.Type + " but this construct is only value for FLAG type fields");
//Old version just set this value
// value = VCFConstants.MISSING_VALUE_v4;
}
else
{
value = true;
}
}
// this line ensures that key/value pairs that look like key=; are parsed correctly as MISSING
if ("".Equals(value))
{
value = VCFConstants.MISSING_VALUE_v4;
}
//TODO: We should have enough information in the header here to
attributes [key] = value;
}
}
return(attributes);
}
/// <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(VCFInfoHeaderLine line)
{
infoStandards.add(line);
}
