private static SAMAlignmentHeader ParseSamHeader(List <string> headerStrings)
{
SAMAlignmentHeader samHeader = new SAMAlignmentHeader();
foreach (string headerString in headerStrings)
{
string[] tokens = headerString.Split(tabDelim, StringSplitOptions.RemoveEmptyEntries);
string recordTypecode = tokens[0].Substring(1);
// Validate the header format.
ValidateHeaderLineFormat(headerString);
SAMRecordField headerLine = null;
if (string.Compare(recordTypecode, "CO", StringComparison.OrdinalIgnoreCase) != 0)
{
List <string> tags = new List <string>();
headerLine = new SAMRecordField(recordTypecode);
for (int i = 1; i < tokens.Length; i++)
{
string tagToken = tokens[i];
string tagName = tagToken.Substring(0, 2);
tags.Add(tagName);
headerLine.Tags.Add(new SAMRecordFieldTag(tagName, tagToken.Substring(3)));
}
samHeader.RecordFields.Add(headerLine);
}
else
{
samHeader.Comments.Add(headerString.Substring(4));
}
}
IList <ReferenceSequenceInfo> referenceSeqsInfo = samHeader.GetReferenceSequencesInfoFromSQHeader();
foreach (var item in referenceSeqsInfo)
{
samHeader.ReferenceSequences.Add(item);
}
string message = samHeader.IsValid();
if (!string.IsNullOrEmpty(message))
{
throw new FormatException(message);
}
return(samHeader);
}
/// <summary>
/// Parses SAM alignment header from specified BioTextReader.
/// </summary>
/// <param name="bioReader">Bio text reader.</param>
private static SAMAlignmentHeader ParserSAMHeader(BioTextReader bioReader)
{
SAMAlignmentHeader samHeader = new SAMAlignmentHeader();
if (bioReader.HasLines && bioReader.Line.StartsWith(@"@", StringComparison.OrdinalIgnoreCase))
{
while (bioReader.HasLines && bioReader.Line.StartsWith(@"@", StringComparison.OrdinalIgnoreCase))
{
string[] tokens = bioReader.Line.Split(tabDelim, StringSplitOptions.RemoveEmptyEntries);
string recordTypecode = tokens[0].Substring(1);
// Validate the header format.
ValidateHeaderLineFormat(bioReader.Line);
SAMRecordField headerLine = null;
if (string.Compare(recordTypecode, "CO", StringComparison.OrdinalIgnoreCase) != 0)
{
List <string> tags = new List <string>();
headerLine = new SAMRecordField(recordTypecode);
for (int i = 1; i < tokens.Length; i++)
{
string tagToken = tokens[i];
string tagName = tagToken.Substring(0, 2);
tags.Add(tagName);
headerLine.Tags.Add(new SAMRecordFieldTag(tagName, tagToken.Substring(3)));
}
samHeader.RecordFields.Add(headerLine);
}
else
{
samHeader.Comments.Add(bioReader.Line.Substring(4));
}
bioReader.GoToNextLine();
}
string message = samHeader.IsValid();
if (!string.IsNullOrEmpty(message))
{
throw new FormatException(message);
}
}
return(samHeader);
}
// compares chromosome name in the specified fields.
private int CompareByChromosomeName(SAMRecordField field1, SAMRecordField field2)
{
string chr1 = field1.Tags.FirstOrDefault(Tag => Tag.Tag.Equals("SN")).Value;
string chr2 = field2.Tags.FirstOrDefault(Tag => Tag.Tag.Equals("SN")).Value;
return string.Compare(chr1, chr2, StringComparison.Ordinal);
}
/// <summary>
/// Add a sequence to the filtered output file header
/// </summary>
private void AddToHeader(SAMAlignedSequence seq)
{
newHeader.ReferenceSequences.Add(new ReferenceSequenceInfo(seq.RName, GetSequence(seq).Length));
// for each good cluster
SAMRecordField sq = new SAMRecordField("SQ");
sq.Tags.Add(new SAMRecordFieldTag("SN", seq.RName));
sq.Tags.Add(new SAMRecordFieldTag("LN", GetSequence(seq).Length.ToString(ci)));
newHeader.RecordFields.Add(sq);
}
