public bool Map()
{
if (MappedAlignment == null)
{
return(false);
}
try
{
rCADDataContext dc = CreateDataContext();
NextSeqID = dc.NextSeqIDs.Select(row => row.SeqID).First();
AlignmentID = dc.NextAlnIDs.Select(row => row.AlnID).First();
NextAlnID = AlignmentID + 1;
AlignmentSeqTypeID = dc.SequenceTypes.Where(row => row.MoleculeType.Equals(_alignment.MoleculeType) && row.GeneName.Equals(_alignment.GeneName) &&
row.GeneType.Equals(_alignment.GeneType)).First().SeqTypeID;
var seqToTaxID = (from seq in _alignment.Sequences
join taxonomyNameRow in dc.TaxonomyNames
on((SequenceMetadata)seq.Metadata[SequenceMetadata.SequenceMetadataLabel]).ScientificName equals taxonomyNameRow.ScientificName
select new { seq.ID, taxonomyNameRow.TaxID }).ToDictionary(match => match.ID, match => match.TaxID);
int rootTaxID = (from taxname in dc.TaxonomyNames
where taxname.ScientificName.Equals("root")
select taxname.TaxID).First();
ExtentTypeIDs = (from bar in dc.SecondaryStructureExtentTypes
select new { bar.ExtentTypeID, bar.ExtentType }).ToDictionary(match => match.ExtentType, match => match.ExtentTypeID);
foreach (var sequence in _alignment.Sequences)
{
SequenceMetadata metadata = (SequenceMetadata)sequence.Metadata[SequenceMetadata.SequenceMetadataLabel];
SequenceMappingData data = new SequenceMappingData();
data.SeqID = NextSeqID;
data.TaxID = seqToTaxID.ContainsKey(sequence.ID) ? seqToTaxID[sequence.ID] : rootTaxID; //The sequence is mapped to the root of the Taxonomy tree if we don't have mapping info.
data.LocationID = dc.CellLocationInfos.Where(row => row.Description.Equals(metadata.LocationDescription)).First().LocationID;
sequence.Metadata.Add(rCADMappingData, data);
NextSeqID++;
}
dc.Connection.Close();
MappedSuccessfully = true;
return(true);
}
catch
{
return(false);
}
}
private void CreateAlignmentImportFiles()
{
StreamWriter alignmentcolumn = File.CreateText(_alignmentColumnFile);
StreamWriter alignment = File.CreateText(_alignmentFile);
StreamWriter alignmentsequence = File.CreateText(_alignmentSequenceDataFile);
StreamWriter alignmentdata = File.CreateText(_alignmentDataFile);
StreamWriter sequencemain = File.CreateText(_sequenceMainFile);
StreamWriter sequenceaccession = File.CreateText(_sequenceAccessionFile);
StreamWriter secondarystructurebasepairs = File.CreateText(_secondaryStructureBasePairsFile);
StreamWriter secondarystructureextents = File.CreateText(_secondaryStructureExtentsFile);
//rCAD.Alignment: Entry in the alignment table for the new alignment
alignment.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}", _data.AlignmentID, _data.AlignmentSeqTypeID, _data.MappedAlignment.LogicalName, _data.MappedAlignment.Columns + 1);
alignment.Flush();
alignment.Close();
//rCAD.AlignmentColumn: We have a 1 to 1 mapping of the logical and physical column numbers at the start
for (int i = 0; i < _data.MappedAlignment.Columns; i++)
{
alignmentcolumn.WriteLine("{0}\t|\t{1}\t|\t{2}", _data.AlignmentID, i + 1, i + 1);
}
alignmentcolumn.Flush();
alignmentcolumn.Close();
//rCAD.AlignmentSequence, rCAD.AlignmentData, rCAD.SequenceMain, rCAD.SequenceAccession written on a per-sequence basis
//We will do duplicate checking inside the database.
foreach (var sequence in _data.MappedAlignment.Sequences)
{
SequenceMetadata metadata = (SequenceMetadata)sequence.Metadata[SequenceMetadata.SequenceMetadataLabel];
SequenceMappingData mappingMetadata = (SequenceMappingData)sequence.Metadata[Mapper.rCADMappingData];
int seqLengthMetadata = metadata.SequenceLength;
int firstNtColNum = -1;
int lastNtColNumber = -1;
int sequenceIndex = 1;
for (int i = 0; i < sequence.Count; i++)
{
if (!sequence[i].IsGap) //We are only actually writing the non-gap positions.
{
alignmentdata.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}", mappingMetadata.SeqID, _data.AlignmentID, i + 1, sequence[i].Symbol, sequenceIndex);
sequenceIndex++;
if (firstNtColNum < 0)
{
firstNtColNum = i + 1;
}
//KJD, 1/21/2010 - A nasty little bug right here where lastNtColNumber = i + 0 SHOULD be lastNtColNumber = i + 1!
if (firstNtColNum > 0)
{
lastNtColNumber = i + 1; //We just set the last col num value to the last column with data we've seen.
}
//lastNtColNumber = i + 0;
}
}
if ((sequenceIndex - 1) != seqLengthMetadata)
{
Console.WriteLine("Warning: Existing metadata for SeqLength ({0}) does not match number of observed nt ({1}) for {2}", seqLengthMetadata, sequenceIndex - 1, sequence.ID);
}
sequencemain.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t\t|\t",
mappingMetadata.SeqID, mappingMetadata.TaxID, mappingMetadata.LocationID, _data.AlignmentSeqTypeID, seqLengthMetadata);
alignmentsequence.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}", mappingMetadata.SeqID, _data.AlignmentID, metadata.AlignmentRowName, firstNtColNum, lastNtColNumber);
foreach (var gbentry in metadata.Accessions)
{
sequenceaccession.WriteLine("{0}\t|\t{1}\t|\t{2}", mappingMetadata.SeqID, gbentry.Accession, gbentry.Version);
}
if (metadata.StructureModel != null && metadata.StructureModel.Pairs.Count() > 0)
{
foreach (int fivePrime in metadata.StructureModel.Pairs.Keys)
{
secondarystructurebasepairs.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}", mappingMetadata.SeqID, _data.AlignmentID, fivePrime,
metadata.StructureModel.Pairs[fivePrime]);
}
int extentID = 1;
foreach (var helix in metadata.StructureModel.Helices)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, helix.FivePrimeStart, helix.FivePrimeEnd, _data.ExtentTypeIDs["Helix"]);
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 2, helix.ThreePrimeStart, helix.ThreePrimeEnd, _data.ExtentTypeIDs["Helix"]);
extentID++;
}
foreach (var hairpinloop in metadata.StructureModel.Hairpins)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, hairpinloop.Loop.LoopStart, hairpinloop.Loop.LoopEnd, _data.ExtentTypeIDs["Hairpin Loop"]);
extentID++;
}
foreach (var internalloop in metadata.StructureModel.Internals)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, internalloop.FivePrimeLoop.LoopStart, internalloop.FivePrimeLoop.LoopEnd, _data.ExtentTypeIDs["Internal Loop"]);
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 2, internalloop.ThreePrimeLoop.LoopStart, internalloop.ThreePrimeLoop.LoopEnd, _data.ExtentTypeIDs["Internal Loop"]);
extentID++;
}
foreach (var bulgeloop in metadata.StructureModel.Bulges)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, bulgeloop.Bulge.LoopStart, bulgeloop.Bulge.LoopEnd, _data.ExtentTypeIDs["Bulge Loop"]);
extentID++;
}
foreach (var multistemloop in metadata.StructureModel.Stems)
{
int stemordinal = 1;
foreach (var stem in multistemloop.Segments)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, stemordinal, stem.LoopStart, stem.LoopEnd, _data.ExtentTypeIDs["Multistem Loop"]);
stemordinal++;
}
extentID++;
}
foreach (var free in metadata.StructureModel.Strands)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, free.LoopStart, free.LoopEnd, _data.ExtentTypeIDs["Free"]);
extentID++;
}
foreach (var tail in metadata.StructureModel.Tails)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, tail.LoopStart, tail.LoopEnd, _data.ExtentTypeIDs["Tail"]);
extentID++;
}
foreach (var knot in metadata.StructureModel.KnottedHelices)
{
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 1, knot.FivePrimeStart, knot.FivePrimeEnd, _data.ExtentTypeIDs["Pseudoknot Helix"]);
secondarystructureextents.WriteLine("{0}\t|\t{1}\t|\t{2}\t|\t{3}\t|\t{4}\t|\t{5}\t|\t{6}", mappingMetadata.SeqID, _data.AlignmentID,
extentID, 2, knot.ThreePrimeStart, knot.ThreePrimeEnd, _data.ExtentTypeIDs["Pseudoknot Helix"]);
extentID++;
}
}
}
sequencemain.Flush();
sequencemain.Close();
alignmentsequence.Flush();
alignmentsequence.Close();
alignmentdata.Flush();
alignmentdata.Close();
sequenceaccession.Flush();
sequenceaccession.Close();
secondarystructurebasepairs.Flush();
secondarystructurebasepairs.Close();
secondarystructureextents.Flush();
secondarystructureextents.Close();
}