/// <summary>
/// jh
/// </summary>
/// <returns></returns>
public SAMAlignedSequence Clone()
{
SAMAlignedSequence n = new SAMAlignedSequence();
n.CIGAR = this.CIGAR;
n.Bin = this.Bin;
n.Flag = this.Flag;
n.ISize = this.ISize;
n.MapQ = this.MapQ;
n.MPos = this.MPos;
n.MRNM = this.MRNM;
n.Pos = this.Pos;
n.QName = this.QName;
n.QuerySequence = new Bio.Sequence(this.QuerySequence.Alphabet, this.QuerySequence.ConvertToString());
n.RName = this.RName;
foreach (SAMOptionalField Item in this.OptionalFields)
{
SAMOptionalField nItem = new SAMOptionalField();
nItem.Tag = Item.Tag;
nItem.Value = Item.Value;
nItem.VType = Item.VType;
n.OptionalFields.Add(nItem);
}
return(n);
}
// parses sequence.
private void ParseSequences(SequenceAlignmentMap seqAlignment, BioTextReader bioReader, bool isReadOnly)
{
while (bioReader.HasLines && !bioReader.Line.StartsWith(@"@", StringComparison.OrdinalIgnoreCase))
{
string[] tokens = bioReader.Line.Split(tabDelim, StringSplitOptions.RemoveEmptyEntries);
SAMAlignedSequence alignedSeq = new SAMAlignedSequence();
alignedSeq.QName = tokens[0];
alignedSeq.Flag = SAMAlignedSequenceHeader.GetFlag(tokens[1]);
alignedSeq.RName = tokens[2];
alignedSeq.Pos = int.Parse(tokens[3], CultureInfo.InvariantCulture);
alignedSeq.MapQ = int.Parse(tokens[4], CultureInfo.InvariantCulture);
alignedSeq.CIGAR = tokens[5];
alignedSeq.MRNM = tokens[6].Equals("=") ? alignedSeq.RName : tokens[6];
alignedSeq.MPos = int.Parse(tokens[7], CultureInfo.InvariantCulture);
alignedSeq.ISize = int.Parse(tokens[8], CultureInfo.InvariantCulture);
string message = alignedSeq.IsValidHeader();
if (!string.IsNullOrEmpty(message))
{
throw new FormatException(message);
}
ISequence refSeq = null;
if (RefSequences != null && RefSequences.Count > 0)
{
refSeq = RefSequences.FirstOrDefault(R => string.Compare(R.ID, alignedSeq.RName, StringComparison.OrdinalIgnoreCase) == 0);
}
ParseQualityNSequence(alignedSeq, Alphabet, Encoding, tokens[9], tokens[10], refSeq, isReadOnly);
SAMOptionalField optField = null;
for (int i = 11; i < tokens.Length; i++)
{
optField = new SAMOptionalField();
string optionalFieldRegExpn = OptionalFieldLinePattern;
if (!Helper.IsValidRegexValue(optionalFieldRegExpn, tokens[i]))
{
message = string.Format(CultureInfo.CurrentCulture, Resource.InvalidOptionalField, tokens[i]);
throw new FormatException(message);
}
string[] opttokens = tokens[i].Split(colonDelim, StringSplitOptions.RemoveEmptyEntries);
optField.Tag = opttokens[0];
optField.VType = opttokens[1];
optField.Value = opttokens[2];
message = optField.IsValid();
if (!string.IsNullOrEmpty(message))
{
throw new FormatException(message);
}
alignedSeq.OptionalFields.Add(optField);
}
seqAlignment.QuerySequences.Add(alignedSeq);
bioReader.GoToNextLine();
}
}
public void AddOptionalField(SAMOptionalField sof)
{
if (UsingReadOnlyOptionFieldCollection)
{
changeFromReadOnlyToEditableCollection();
}
else if (editableOptionalFieldCollection == null)
{
editableOptionalFieldCollection = new Dictionary <string, SAMOptionalField>();
}
editableOptionalFieldCollection[sof.Tag] = sof;
}
/// <summary>
/// Convert this read only collection into a dictionary of option fields
/// </summary>
/// <returns></returns>
public Dictionary <string, SAMOptionalField> ConvertToDictionary()
{
Dictionary <string, SAMOptionalField> toReturn = new Dictionary <string, SAMOptionalField>(tagToDataTypeAndLocation.Count);
foreach (var kv in tagToDataTypeAndLocation)
{
char type = kv.Value.Key;
object value = this[kv.Key].ToString();
string tag = kv.Key;
SAMOptionalField sof = new SAMOptionalField(tag, value, type);
toReturn[tag] = sof;
}
return(toReturn);
}
/// <summary>
/// Parse a single sequencer.
/// </summary>
/// <param name="bioText">sequence alignment text.</param>
/// <param name="alphabet">Alphabet of the sequences.</param>
/// <param name="referenceSequences">Reference sequences.</param>
private static SAMAlignedSequence ParseSequence(string bioText, IAlphabet alphabet, IList <ISequence> referenceSequences)
{
const int optionalTokenStartingIndex = 11;
string[] tokens = bioText.Split(tabDelim, StringSplitOptions.RemoveEmptyEntries);
SAMAlignedSequence alignedSeq = new SAMAlignedSequence();
alignedSeq.QName = tokens[0];
alignedSeq.Flag = SAMAlignedSequenceHeader.GetFlag(tokens[1]);
alignedSeq.RName = tokens[2];
alignedSeq.Pos = int.Parse(tokens[3], CultureInfo.InvariantCulture);
alignedSeq.MapQ = int.Parse(tokens[4], CultureInfo.InvariantCulture);
alignedSeq.CIGAR = tokens[5];
alignedSeq.MRNM = tokens[6].Equals("=") ? alignedSeq.RName : tokens[6];
alignedSeq.MPos = int.Parse(tokens[7], CultureInfo.InvariantCulture);
alignedSeq.ISize = int.Parse(tokens[8], CultureInfo.InvariantCulture);
ISequence refSeq = null;
if (referenceSequences != null && referenceSequences.Count > 0)
{
refSeq = referenceSequences.FirstOrDefault(R => string.Compare(R.ID, alignedSeq.RName, StringComparison.OrdinalIgnoreCase) == 0);
}
ParseQualityNSequence(alignedSeq, alphabet, tokens[9], tokens[10], refSeq);
SAMOptionalField optField = null;
string message;
for (int i = optionalTokenStartingIndex; i < tokens.Length; i++)
{
optField = new SAMOptionalField();
if (!Helper.IsValidRegexValue(OptionalFieldRegex, tokens[i]))
{
message = string.Format(CultureInfo.CurrentCulture, Properties.Resource.InvalidOptionalField, tokens[i]);
throw new FormatException(message);
}
string[] opttokens = tokens[i].Split(colonDelim, StringSplitOptions.RemoveEmptyEntries);
optField.Tag = opttokens[0];
optField.VType = opttokens[1];
optField.Value = opttokens[2];
alignedSeq.OptionalFields.Add(optField);
}
return(alignedSeq);
}
/// <summary>
/// Parse a single sequencer.
/// </summary>
/// <param name="bioText">sequence alignment text.</param>
/// <param name="alphabet">Alphabet of the sequences.</param>
public static SAMAlignedSequence ParseSequence(string bioText, IAlphabet alphabet)
{
const int optionalTokenStartingIndex = 11;
string[] tokens = bioText.Split(TabDelim, StringSplitOptions.RemoveEmptyEntries);
SAMAlignedSequence alignedSeq = new SAMAlignedSequence
{
QName = tokens[0],
Flag = SAMAlignedSequenceHeader.GetFlag(tokens[1]),
RName = tokens[2],
Pos = int.Parse(tokens[3]),
MapQ = int.Parse(tokens[4]),
CIGAR = tokens[5]
};
alignedSeq.MRNM = tokens[6].Equals("=") ? alignedSeq.RName : tokens[6];
alignedSeq.MPos = int.Parse(tokens[7]);
alignedSeq.ISize = int.Parse(tokens[8]);
ParseQualityNSequence(alignedSeq, alphabet, tokens[9], tokens[10]);
for (int i = optionalTokenStartingIndex; i < tokens.Length; i++)
{
SAMOptionalField optField = new SAMOptionalField();
if (!Helper.IsValidRegexValue(OptionalFieldRegex, tokens[i]))
{
throw new FormatException(string.Format(Properties.Resource.InvalidOptionalField, tokens[i]));
}
string[] opttokens = tokens[i].Split(ColonDelim, StringSplitOptions.RemoveEmptyEntries);
optField.Tag = opttokens[0];
optField.VType = opttokens[1];
optField.Value = opttokens[2];
alignedSeq.OptionalFields.Add(optField);
}
return(alignedSeq);
}
// Gets optional field in a byte array.
private static byte[] GetOptioanField(SAMOptionalField field)
{
int valueSize = GetOptionalFieldValueSize(field);
if (valueSize == 0)
{
string message = string.Format(CultureInfo.InvariantCulture, Properties.Resource.BAM_InvalidIntValueInOptField, field.Value, field.Tag);
throw new FormatException(message);
}
int arrayLen = valueSize < 0 ? -valueSize : valueSize;
arrayLen += 3;
byte[] array = new byte[arrayLen];
array[0] = (byte)field.Tag[0];
array[1] = (byte)field.Tag[1];
array[2] = (byte)field.VType[0];
byte[] temparray = new byte[4];
switch (field.VType)
{
case "A": // Printable character
array[3] = (byte)field.Value[0];
break;
case "c": //signed 8-bit integer
case "C": //unsigned 8-bit integer
case "s": // signed 16 bit integer
case "S"://unsinged 16 bit integer
case "i": // signed 32 bit integer
case "I": // unsigned 32 bit integer
if (valueSize == 1)
{
array[2] = (byte)'C';
array[3] = byte.Parse(field.Value, CultureInfo.InvariantCulture);
}
else if (valueSize == -1)
{
sbyte sb = sbyte.Parse(field.Value, CultureInfo.InvariantCulture);
array[2] = (byte)'c';
array[3] = (byte)sb;
}
else if (valueSize == 2)
{
UInt16 uint16value = UInt16.Parse(field.Value, CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(uint16value);
array[2] = (byte)'S';
array[3] = temparray[0];
array[4] = temparray[1];
}
else if (valueSize == -2)
{
Int16 int16value = Int16.Parse(field.Value, CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(int16value);
array[2] = (byte)'s';
array[3] = temparray[0];
array[4] = temparray[1];
}
else if (valueSize == 4)
{
uint uint32value = uint.Parse(field.Value, CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(uint32value);
array[2] = (byte)'I';
array[3] = temparray[0];
array[4] = temparray[1];
array[5] = temparray[2];
array[6] = temparray[3];
}
else
{
int int32value = int.Parse(field.Value, CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(int32value);
array[2] = (byte)'i';
array[3] = temparray[0];
array[4] = temparray[1];
array[5] = temparray[2];
array[6] = temparray[3];
}
break;
case "f": // float
float floatvalue = float.Parse(field.Value, CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(floatvalue);
array[3] = temparray[0];
array[4] = temparray[1];
array[5] = temparray[2];
array[6] = temparray[3];
break;
case "Z": // printable string
temparray = Encoding.UTF8.GetBytes(field.Value);
temparray.CopyTo(array, 3);
array[3 + temparray.Length] = (byte)'\0';
break;
case "H": // HexString
temparray = Encoding.UTF8.GetBytes(field.Value);
temparray.CopyTo(array, 3);
array[3 + temparray.Length] = (byte)'\0';
break;
case "B": // integer or numeric array.
UpdateArrayType(array, field);
break;
default:
throw new Exception(Properties.Resource.BAM_InvalidOptValType);
}
return array;
}
// Gets optional field value size.
private static int GetOptionalFieldValueSize(SAMOptionalField optionalField)
{
switch (optionalField.VType)
{
case "A": // Printable character
case "c": //signed 8-bit integer
return -1;
case "C": //unsigned 8-bit integer
return 1;
case "s": // signed 16 bit integer
case "S"://unsinged 16 bit integer
case "i": // signed 32 bit integer
case "I": // unsigned 32 bit integer
return GetOptionalFieldIntValueSize(optionalField.Value);
case "f": // float
return 4;
case "Z": // printable string
case "H": // HexString
return optionalField.Value.Length + 1;
case "B"://integer or numeric array
char type = optionalField.Value[0];
int arrayTypeSize = GetSizeOfArrayType(type);
int numberofelements = optionalField.Value.Split(DelimComma, StringSplitOptions.RemoveEmptyEntries).Length - 1;
int elementsSize = arrayTypeSize * numberofelements;
int arraylen = elementsSize + 1 + 4; // 1 to store array type and 4 to store number of values in array.
return arraylen;
default:
throw new Exception(Properties.Resource.BAM_InvalidOptValType);
}
}
private static void UpdateArrayType(byte[] array, SAMOptionalField field)
{
byte[] temparray = new byte[4];
char arraytype = field.Value[0];
int arrayTypeSize = GetSizeOfArrayType(arraytype);
string[] elements = field.Value.Split(DelimComma, StringSplitOptions.RemoveEmptyEntries);
array[3] = (byte)arraytype;
int arrayIndex = 4;
temparray = Helper.GetLittleEndianByteArray(elements.Length - 1);
array[arrayIndex++] = temparray[0];
array[arrayIndex++] = temparray[1];
array[arrayIndex++] = temparray[2];
array[arrayIndex++] = temparray[3];
//elemetns[0] contains array type;
for (int i = 1; i < elements.Length; i++)
{
switch (arraytype)
{
case 'A': // Printable character
temparray[0] = (byte)elements[i][0];
break;
case 'c': //signed 8-bit integer
temparray[0] = (byte)sbyte.Parse(elements[i], CultureInfo.InvariantCulture);
break;
case 'C': //unsigned 8-bit integer
temparray[0] = byte.Parse(elements[i], CultureInfo.InvariantCulture);
break;
case 's': // signed 16 bit integer
Int16 int16value = Int16.Parse(elements[i], CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(int16value);
break;
case 'S'://unsinged 16 bit integer
UInt16 uint16value = UInt16.Parse(elements[i], CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(uint16value);
break;
case 'i': // signed 32 bit integer
int int32value = int.Parse(elements[i], CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(int32value);
break;
case 'I': // unsigned 32 bit integer
uint uint32value = uint.Parse(elements[i], CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(uint32value);
break;
case 'f': // float
float floatvalue = float.Parse(elements[i], CultureInfo.InvariantCulture);
temparray = Helper.GetLittleEndianByteArray(floatvalue);
break;
default:
throw new Exception(string.Format(Properties.Resource.BAM_InvalidOptValType, arraytype));
}
for (int tempIndex = 0; tempIndex < arrayTypeSize; tempIndex++)
{
array[arrayIndex++] = temparray[tempIndex];
}
}
}
/// <summary>
/// Returns an aligned sequence by parses the BAM file.
/// </summary>
private SAMAlignedSequence GetAlignedSequence(int start, int end)
{
byte[] array = new byte[4];
ReadUnCompressedData(array, 0, 4);
int blockLen = Helper.GetInt32(array, 0);
byte[] alignmentBlock = new byte[blockLen];
ReadUnCompressedData(alignmentBlock, 0, blockLen);
SAMAlignedSequence alignedSeq = new SAMAlignedSequence();
int value;
UInt32 UnsignedValue;
// 0-4 bytes
int refSeqIndex = Helper.GetInt32(alignmentBlock, 0);
if (refSeqIndex == -1)
alignedSeq.RName = "*";
else
alignedSeq.RName = refSeqNames[refSeqIndex];
// 4-8 bytes
alignedSeq.Pos = Helper.GetInt32(alignmentBlock, 4) + 1;
// if there is no overlap no need to parse further.
// BAMPos > closedEnd
// => (alignedSeq.Pos - 1) > end -1
if (alignedSeq.Pos > end)
{
return null;
}
// 8 - 12 bytes "bin<<16|mapQual<<8|read_name_len"
UnsignedValue = Helper.GetUInt32(alignmentBlock, 8);
// 10 -12 bytes
alignedSeq.Bin = (int)(UnsignedValue & 0xFFFF0000) >> 16;
// 9th bytes
alignedSeq.MapQ = (int)(UnsignedValue & 0x0000FF00) >> 8;
// 8th bytes
int queryNameLen = (int)(UnsignedValue & 0x000000FF);
// 12 - 16 bytes
UnsignedValue = Helper.GetUInt32(alignmentBlock, 12);
// 14-16 bytes
int flagValue = (int)(UnsignedValue & 0xFFFF0000) >> 16;
alignedSeq.Flag = (SAMFlags)flagValue;
// 12-14 bytes
int cigarLen = (int)(UnsignedValue & 0x0000FFFF);
// 16-20 bytes
int readLen = Helper.GetInt32(alignmentBlock, 16);
// 20-24 bytes
int mateRefSeqIndex = Helper.GetInt32(alignmentBlock, 20);
if (mateRefSeqIndex != -1)
{
alignedSeq.MRNM = refSeqNames[mateRefSeqIndex];
}
else
{
alignedSeq.MRNM = "*";
}
// 24-28 bytes
alignedSeq.MPos = Helper.GetInt32(alignmentBlock, 24) + 1;
// 28-32 bytes
alignedSeq.ISize = Helper.GetInt32(alignmentBlock, 28);
// 32-(32+readLen) bytes
alignedSeq.QName = System.Text.ASCIIEncoding.ASCII.GetString(alignmentBlock, 32, queryNameLen - 1);
StringBuilder strbuilder = new StringBuilder();
int startIndex = 32 + queryNameLen;
for (int i = startIndex; i < (startIndex + cigarLen * 4); i += 4)
{
// Get the CIGAR operation length stored in first 28 bits.
UInt32 cigarValue = Helper.GetUInt32(alignmentBlock, i);
strbuilder.Append(((cigarValue & 0xFFFFFFF0) >> 4).ToString(CultureInfo.InvariantCulture));
// Get the CIGAR operation stored in last 4 bits.
value = (int)cigarValue & 0x0000000F;
// MIDNSHP=>0123456
switch (value)
{
case 0:
strbuilder.Append("M");
break;
case 1:
strbuilder.Append("I");
break;
case 2:
strbuilder.Append("D");
break;
case 3:
strbuilder.Append("N");
break;
case 4:
strbuilder.Append("S");
break;
case 5:
strbuilder.Append("H");
break;
case 6:
strbuilder.Append("P");
break;
case 7:
strbuilder.Append("=");
break;
case 8:
strbuilder.Append("X");
break;
default:
throw new FileFormatException(Properties.Resource.BAM_InvalidCIGAR);
}
}
string cigar = strbuilder.ToString();
if (string.IsNullOrWhiteSpace(cigar))
{
alignedSeq.CIGAR = "*";
}
else
{
alignedSeq.CIGAR = cigar;
}
// if there is no overlap no need to parse further.
// ZeroBasedRefEnd < start
// => (alignedSeq.RefEndPos -1) < start
if (alignedSeq.RefEndPos - 1 < start && alignedSeq.RName!=Properties.Resource.SAM_NO_REFERENCE_DEFINED_INDICATOR)
{
return null;
}
startIndex += cigarLen * 4;
strbuilder = new StringBuilder();
int index = startIndex;
for (; index < (startIndex + (readLen + 1) / 2) - 1; index++)
{
// Get first 4 bit value
value = (alignmentBlock[index] & 0xF0) >> 4;
strbuilder.Append(GetSeqChar(value));
// Get last 4 bit value
value = alignmentBlock[index] & 0x0F;
strbuilder.Append(GetSeqChar(value));
}
value = (alignmentBlock[index] & 0xF0) >> 4;
strbuilder.Append(GetSeqChar(value));
if (readLen % 2 == 0)
{
value = alignmentBlock[index] & 0x0F;
strbuilder.Append(GetSeqChar(value));
}
startIndex = index + 1;
string strSequence = strbuilder.ToString();
byte[] qualValues = new byte[readLen];
string strQualValues = "*";
if (alignmentBlock[startIndex] != 0xFF)
{
for (int i = startIndex; i < (startIndex + readLen); i++)
{
qualValues[i - startIndex] = (byte)(alignmentBlock[i] + 33);
}
strQualValues = System.Text.ASCIIEncoding.ASCII.GetString(qualValues);
}
SAMParser.ParseQualityNSequence(alignedSeq, Alphabet, strSequence, strQualValues);
startIndex += readLen;
if (alignmentBlock.Length > startIndex + 4 && alignmentBlock[startIndex] != 0x0 && alignmentBlock[startIndex + 1] != 0x0)
{
for (index = startIndex; index < alignmentBlock.Length; )
{
SAMOptionalField optionalField = new SAMOptionalField();
optionalField.Tag = System.Text.ASCIIEncoding.ASCII.GetString(alignmentBlock, index, 2);
index += 2;
char vType = (char)alignmentBlock[index++];
string valueType = vType.ToString();
// SAM format supports [AifZH] for value type.
// In BAM, an integer may be stored as a signed 8-bit integer (c), unsigned 8-bit integer (C), signed short (s), unsigned
// short (S), signed 32-bit (i) or unsigned 32-bit integer (I), depending on the signed magnitude of the integer. However,
// in SAM, all types of integers are presented as type ʻiʼ.
string message = Helper.IsValidPatternValue("VType", valueType, BAMOptionalFieldRegex);
if (!string.IsNullOrEmpty(message))
{
throw new FormatException(message);
}
optionalField.Value = GetOptionalValue(vType, alignmentBlock, ref index).ToString();
// Convert to SAM format.
if ("cCsSI".IndexOf(vType) >= 0)
{
valueType = "i";
}
optionalField.VType = valueType;
alignedSeq.OptionalFields.Add(optionalField);
}
}
return alignedSeq;
}