private static IGenomesReferencePath SafeGetReference(string fastaPath)
{
IGenomesReferencePath reference = null;
try
{
reference = new IGenomesReferencePath(fastaPath);
}
catch { }
return(reference);
}
public static bool SameReference(string fastaA, string fastaB)
{
IGenomesReferencePath fastaAReference = null;
IGenomesReferencePath fastaBReference = null;
try
{
fastaAReference = new IGenomesReferencePath(fastaA);
}
catch (ArgumentException)
{ }
try
{
fastaBReference = new IGenomesReferencePath(fastaB);
}
catch (ArgumentException)
{ }
if (fastaAReference == null || fastaBReference == null)
{
return(fastaA == fastaB);
}
return(fastaAReference.Equals(fastaBReference));
}
/// <summary>
/// Populates the genome metadata from an XML file
/// </summary>
public void Deserialize(string inputFilename)
{
// open the XML file
inputFilename = Path.GetFullPath(inputFilename);
string directory = Path.GetDirectoryName(inputFilename);
Length = 0;
KnownBases = 0; // initial
int refIndex = 0;
IGenomesReferencePath iGenomesReference = IGenomesReferencePath.GetReferenceFromFastaPath(directory);
// use StreamReader to avoid URI parsing of filename that will cause problems with
// certain characters in the path (#).
using (var xmlReader = XmlReader.Create(new StreamReader(inputFilename)))
{
while (xmlReader.Read())
{
XmlNodeType nType = xmlReader.NodeType;
// handle
if (nType == XmlNodeType.Element)
{
// retrieve the genome variables
if (xmlReader.Name == "sequenceSizes")
{
Name = xmlReader.GetAttribute("genomeName");
if (iGenomesReference != null && string.IsNullOrEmpty(Name))
{
Name = iGenomesReference.ToString();
}
}
// retrieve the chromosome variables
if (xmlReader.Name == "chromosome")
{
SequenceMetadata refSeq = new SequenceMetadata
{
FastaPath = Path.Combine(directory, xmlReader.GetAttribute("fileName")),
Name = xmlReader.GetAttribute("contigName"),
Index = refIndex++,
Length = long.Parse(xmlReader.GetAttribute("totalBases")),
Type = ParseSequenceType(xmlReader.GetAttribute("type"))
};
Length += refSeq.Length;
refSeq.Build = xmlReader.GetAttribute("build");
refSeq.Species = xmlReader.GetAttribute("species");
// update species and build from fasta path if in iGenomes format
if (iGenomesReference != null)
{
if (string.IsNullOrEmpty(refSeq.Build))
{
refSeq.Build = iGenomesReference.Build;
}
if (string.IsNullOrEmpty(refSeq.Species))
{
refSeq.Species = iGenomesReference.Species;
}
}
string isCircular = xmlReader.GetAttribute("isCircular");
if (!string.IsNullOrEmpty(isCircular))
{
refSeq.IsCircular = (isCircular == "true");
}
string ploidy = xmlReader.GetAttribute("ploidy");
if (!string.IsNullOrEmpty(ploidy))
{
refSeq.Ploidy = int.Parse(ploidy);
}
string md5 = xmlReader.GetAttribute("md5");
if (!string.IsNullOrEmpty(md5))
{
refSeq.Checksum = md5;
}
string knownBases = xmlReader.GetAttribute("knownBases");
if (!string.IsNullOrEmpty(knownBases))
{
refSeq.KnownBases = long.Parse(knownBases);
KnownBases += refSeq.KnownBases;
}
Sequences.Add(refSeq);
}
}
}
}
}
// TODO - If anyone is going to use this, put it somewhere appropriate. Probably in BamReader.
//public static IGenomesReferencePath GetReferenceFromBamHeader(string bamPath)
//{
// using (BamReader reader = new BamReader(bamPath))
// {
// return SafeGetReference(reader.GetReferenceURI());
// }
//}
public bool Equals(IGenomesReferencePath p)
{
return(p.Species == Species && p.Provider == Provider && p.Build == Build);
}
