/// <summary>
/// create a VCF format header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the VCF header version </param>
/// <param name="lineType"> the header line type
/// </param>
protected internal VCFCompoundHeaderLine(string line, VCFHeaderVersion version, SupportedHeaderLineType lineType) : base(lineType.ToString(), "")
{
IDictionary <string, string> mapping = VCFHeaderLineTranslator.parseLine(version, line, "ID", "Number", "Type", "Description");
name = mapping["ID"];
count = -1;
string numberStr = mapping["Number"];
if (numberStr.Equals(VCFConstants.PER_ALTERNATE_ALLELE_COUNT))
{
countType = VCFHeaderLineCount.A;
}
else if (numberStr.Equals(VCFConstants.PER_ALLELE_COUNT))
{
countType = VCFHeaderLineCount.R;
}
else if (numberStr.Equals(VCFConstants.PER_GENOTYPE_COUNT))
{
countType = VCFHeaderLineCount.G;
}
else if (((version == VCFHeaderVersion.VCF4_0 || version == VCFHeaderVersion.VCF4_1 || version == VCFHeaderVersion.VCF4_2) && numberStr.Equals(VCFConstants.UNBOUNDED_ENCODING_v4)) || ((version == VCFHeaderVersion.VCF3_2 || version == VCFHeaderVersion.VCF3_3) && numberStr.Equals(VCFConstants.UNBOUNDED_ENCODING_v3)))
{
countType = VCFHeaderLineCount.UNBOUNDED;
}
else
{
countType = VCFHeaderLineCount.INTEGER;
count = Convert.ToInt32(numberStr);
}
if (count < 0 && countType == VCFHeaderLineCount.INTEGER)
{
throw new VCFParsingError("Count < 0 for fixed size VCF header field " + name);
}
try
{
type = (VCFHeaderLineType)Enum.Parse(typeof(VCFHeaderLineType), mapping["Type"]);
}
#pragma warning disable 0168
catch (Exception e)
#pragma warning restore 0168
{
throw new VCFParsingError(mapping["Type"] + " is not a valid type in the VCF specification (note that types are case-sensitive)");
}
if (type == VCFHeaderLineType.Flag && !allowFlagValues())
{
throw new ArgumentException("Flag is an unsupported type for this kind of field");
}
description = mapping["Description"];
if (description == null && ALLOW_UNBOUND_DESCRIPTIONS) // handle the case where there's no description provided
{
description = UNBOUND_DESCRIPTION;
}
this.lineType = lineType;
validate();
}
void writeHeader(VCFHeader header)
{
header = doNotWriteGenotypes ? new VCFHeader(header.MetaDataInSortedOrder) : header;
try
{
// the file format field needs to be written first
writer.Write(VERSION_LINE + "\n");
foreach (VCFHeaderLine line in header.MetaDataInSortedOrder)
{
if (VCFHeaderVersion.IsFormatString(line.Key))
{
continue;
}
writer.Write(VCFHeader.METADATA_INDICATOR);
writer.Write(line.ToString());
writer.Write("\n");
}
// write out the column line
writer.Write(VCFHeader.HEADER_INDICATOR);
bool isFirst = true;
foreach (string field in VCFHeader.HEADER_FIELDS)
{
if (isFirst)
{
isFirst = false; // don't write out a field separator
}
else
{
writer.Write(VCFConstants.FIELD_SEPARATOR);
}
writer.Write(field.ToString());
}
if (header.hasGenotypingData())
{
writer.Write(VCFConstants.FIELD_SEPARATOR);
writer.Write("FORMAT");
foreach (string sample in header.GenotypeSampleNames)
{
writer.Write(VCFConstants.FIELD_SEPARATOR);
writer.Write(sample);
}
}
writer.Write("\n");
}
catch (IOException e)
{
throw new Exception("IOException writing the VCF header.", e);
}
}
/// <summary>
/// check our metadata for a VCF version tag, and throw an exception if the version is out of date
/// or the version is not present
/// TODO: Should only be one format line
/// </summary>
public virtual void loadVCFVersion()
{
IList <VCFHeaderLine> toRemove = new List <VCFHeaderLine>();
foreach (VCFHeaderLine line in mMetaData)
{
if (VCFHeaderVersion.IsFormatString(line.Key))
{
toRemove.Add(line);
}
}
// remove old header lines for now,
mMetaData.RemoveRange(toRemove);
}
/// <param name="reader"> the line reader to take header lines from </param>
/// <returns> the number of header lines </returns>
public virtual VCFHeader readHeader(StreamReader reader)
{
IList <string> headerStrings = new List <string>();
string line;
try
{
bool foundHeaderVersion = false;
while ((line = reader.ReadLine()) != null)
{
lineNo++;
if (line.StartsWith(VCFHeader.METADATA_INDICATOR))
{
string[] lineFields = line.Substring(2).Split('=');
if (lineFields.Length == 2 && VCFHeaderVersion.IsFormatString(lineFields[0]))
{
if (!VCFHeaderVersion.IsVersionString(lineFields[1]))
{
throw new VCFParsingError("Header: " + lineFields[1] + " is not a supported version");
}
foundHeaderVersion = true;
version = VCFHeaderVersion.ToHeaderVersion(lineFields[1]);
if (!this.AcceptableVersions.Contains(version))
{
throw new VCFParsingError("This codec is strictly for " + Name + "; please use a different codec for " + lineFields[1]);
}
}
headerStrings.Add(line);
}
else if (line.StartsWith(VCFHeader.HEADER_INDICATOR))
{//should be only one such line
if (!foundHeaderVersion)
{
throw new VCFParsingError("We never saw a header line specifying VCF version");
}
headerStrings.Add(line);
return(parseHeaderFromLines(headerStrings, version));
}
else
{
throw new VCFParsingError("We never saw the required CHROM header line (starting with one #) for the input VCF file");
}
}
}
catch (IOException e)
{
throw new Exception("IO Exception ", e);
}
throw new VCFParsingError("We never saw the required CHROM header line (starting with one #) for the input VCF file");
}
public static VCFHeaderVersion GetHeaderVersion(String versionLine)
{
versionLine = clean(versionLine);
String[] lineFields = versionLine.Split('=');
if (lineFields.Length != 2 || !IsFormatString(lineFields[0].Substring(2)))
{
throw new VCFParsingError(versionLine + " is not a valid VCF version line");
}
VCFHeaderVersion vcfHV = ToHeaderVersion(lineFields[1]);
if (vcfHV == null)
{
throw new VCFParsingError(lineFields[1] + " is not a supported version");
}
return(vcfHV);
}
public static IDictionary<string, string> parseLine(VCFHeaderVersion version, string valueLine, params string[] expectedTagOrder)
{
return mapping[version].parseLine(valueLine,expectedTagOrder);
}
/// <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 format header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the VCF header version </param>
/// <param name="lineType"> the header line type
/// </param>
protected internal VCFCompoundHeaderLine(string line, VCFHeaderVersion version, SupportedHeaderLineType lineType) : base(lineType.ToString(), "")
{
IDictionary<string, string> mapping = VCFHeaderLineTranslator.parseLine(version,line, "ID","Number","Type","Description");
name = mapping["ID"];
count = -1;
string numberStr = mapping["Number"];
if (numberStr.Equals(VCFConstants.PER_ALLELE_COUNT))
{
countType = VCFHeaderLineCount.A;
}
else if (numberStr.Equals(VCFConstants.PER_GENOTYPE_COUNT))
{
countType = VCFHeaderLineCount.G;
}
else if (((version == VCFHeaderVersion.VCF4_0 || version == VCFHeaderVersion.VCF4_1 || version == VCFHeaderVersion.VCF4_2) && numberStr.Equals(VCFConstants.UNBOUNDED_ENCODING_v4)) || ((version == VCFHeaderVersion.VCF3_2 || version == VCFHeaderVersion.VCF3_3) && numberStr.Equals(VCFConstants.UNBOUNDED_ENCODING_v3)))
{
countType = VCFHeaderLineCount.UNBOUNDED;
}
else
{
countType = VCFHeaderLineCount.INTEGER;
count = Convert.ToInt32(numberStr);
}
if (count < 0 && countType == VCFHeaderLineCount.INTEGER)
{
throw new VCFParsingError("Count < 0 for fixed size VCF header field " + name);
}
try
{
type = (VCFHeaderLineType) Enum.Parse(typeof(VCFHeaderLineType), mapping["Type"]);
}
catch (Exception e)
{
throw new VCFParsingError(mapping["Type"] + " is not a valid type in the VCF specification (note that types are case-sensitive)");
}
if (type == VCFHeaderLineType.Flag && !allowFlagValues())
{
throw new System.ArgumentException("Flag is an unsupported type for this kind of field");
}
description = mapping["Description"];
if (description == null && ALLOW_UNBOUND_DESCRIPTIONS) // handle the case where there's no description provided
{
description = UNBOUND_DESCRIPTION;
}
this.lineType = lineType;
validate();
}
/// <summary>
/// create a VCF info header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the vcf header version </param>
public VCFFilterHeaderLine(string line, VCFHeaderVersion version) : base(line, version, "FILTER", "ID", "Description")
{
}
/// <param name="reader"> the line reader to take header lines from </param>
/// <returns> the number of header lines </returns>
public virtual VCFHeader readHeader (StreamReader reader)
{
IList<string> headerStrings = new List<string> ();
string line;
try {
bool foundHeaderVersion = false;
while ((line = reader.ReadLine ()) != null) {
lineNo++;
if (line.StartsWith (VCFHeader.METADATA_INDICATOR)) {
string[] lineFields = line.Substring (2).Split ('=');
if (lineFields.Length == 2 && VCFHeaderVersion.IsFormatString (lineFields [0])) {
if (!VCFHeaderVersion.IsVersionString (lineFields [1])) {
throw new VCFParsingError ("Header: " + lineFields [1] + " is not a supported version");
}
foundHeaderVersion = true;
version = VCFHeaderVersion.ToHeaderVersion (lineFields [1]);
if (!this.AcceptableVersions.Contains (version)) {
throw new VCFParsingError ("This codec is strictly for " + Name + "; please use a different codec for " + lineFields [1]);
}
}
headerStrings.Add (line);
} else if (line.StartsWith (VCFHeader.HEADER_INDICATOR)) {//should be only one such line
if (!foundHeaderVersion) {
throw new VCFParsingError ("We never saw a header line specifying VCF version");
}
headerStrings.Add (line);
return parseHeaderFromLines (headerStrings, version);
} else {
throw new VCFParsingError ("We never saw the required CHROM header line (starting with one #) for the input VCF file");
}
}
} catch (IOException e) {
throw new Exception ("IO Exception ", e);
}
throw new VCFParsingError ("We never saw the required CHROM header line (starting with one #) for the input VCF file");
}
/// <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;
}
public VCFFormatHeaderLine(string line, VCFHeaderVersion version)
: base(line, version, SupportedHeaderLineType.FORMAT)
{
}
/// <summary>
/// create a VCF info header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the vcf header version </param>
/// <param name="key"> the key for this header line </param>
/// <param name="expectedTagOrdering"> the tag ordering expected for this header line </param>
public VCFSimpleHeaderLine(string line, VCFHeaderVersion version, string key, params string[] expectedTagOrdering)
: this(key, VCFHeaderLineTranslator.parseLine(version, line, expectedTagOrdering), expectedTagOrdering)
{
}
/// <summary>
/// create a VCF contig header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the vcf header version </param>
/// <param name="key"> the key for this header line </param>
//JAVA TO C# CONVERTER WARNING: 'final' parameters are not allowed in .NET:
//ORIGINAL LINE: public VCFContigHeaderLine(final String line, final VCFHeaderVersion version, final String key, int contigIndex)
public VCFContigHeaderLine(string line, VCFHeaderVersion version, string key, int contigIndex) : base(line, version, key, null)
{
this.contigIndex = contigIndex;
}
/// <summary>
/// create a VCF info header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the vcf header version </param>
public VCFFilterHeaderLine(string line, VCFHeaderVersion version) : base(line, version, "FILTER", "ID", "Description")
{
}
/// <summary>
/// create a VCF info header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the vcf header version </param>
/// <param name="key"> the key for this header line </param>
/// <param name="expectedTagOrdering"> the tag ordering expected for this header line </param>
public VCFSimpleHeaderLine(string line, VCFHeaderVersion version, string key, params string[] expectedTagOrdering)
: this(key, VCFHeaderLineTranslator.parseLine(version, line, expectedTagOrdering), expectedTagOrdering)
{
}
/// <summary>
/// create a VCF contig header line
/// </summary>
/// <param name="line"> the header line </param>
/// <param name="version"> the vcf header version </param>
/// <param name="key"> the key for this header line </param>
//JAVA TO C# CONVERTER WARNING: 'final' parameters are not allowed in .NET:
//ORIGINAL LINE: public VCFContigHeaderLine(final String line, final VCFHeaderVersion version, final String key, int contigIndex)
public VCFContigHeaderLine(string line, VCFHeaderVersion version, string key, int contigIndex) : base(line, version, key, null)
{
this.contigIndex = contigIndex;
}
public static IDictionary <string, string> parseLine(VCFHeaderVersion version, string valueLine, params string[] expectedTagOrder)
{
return(mapping[version].parseLine(valueLine, expectedTagOrder));
}
