/// <summary>
/// returns an array of exons given a list of ObjectValues (AbstractData)
/// </summary>
public static MutableExon[] ParseList(IImportNode importNode, IChromosome chromosome)
{
var listMembers = importNode.GetListMembers();
if (listMembers == null)
{
throw new InvalidDataException("Encountered an exon node that could not be converted to a member list.");
}
var exons = new MutableExon[listMembers.Count];
for (var exonIndex = 0; exonIndex < listMembers.Count; exonIndex++)
{
if (listMembers[exonIndex] is ObjectValueNode objectValue)
{
exons[exonIndex] = Parse(objectValue, chromosome);
}
else
{
throw new InvalidDataException($"Could not transform the AbstractData object into an ObjectValue: [{listMembers[exonIndex].GetType()}]");
}
}
return(exons);
}
private MutableExon[] ReadExons(IChromosome chromosome)
{
var cols = GetColumns("Exons");
int numExons = int.Parse(cols[1]);
if (numExons == 0)
{
return(null);
}
var exons = new MutableExon[numExons];
var colIndex = 2;
for (var i = 0; i < numExons; i++)
{
int start = int.Parse(cols[colIndex++]);
int end = int.Parse(cols[colIndex++]);
var phase = (byte)(int.Parse(cols[colIndex++]) + 1);
exons[i] = new MutableExon(chromosome, start, end, phase);
}
return(exons);
}
endExon) Parse(IImportNode importNode, IChromosome currentChromosome)
{
var objectValue = importNode.GetObjectValueNode();
if (objectValue == null)
{
throw new InvalidDataException("Encountered a translation import node that could not be converted to an object value node.");
}
int start = -1;
int end = -1;
string proteinId = null;
byte proteinVersion = 0;
MutableExon startExon = null;
MutableExon endExon = null;
foreach (var node in objectValue.Values)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(node.Key))
{
throw new InvalidDataException($"Encountered an unknown key in the dumper mapper object: {node.Key}");
}
ObjectKeyValueNode exonNode;
switch (node.Key)
{
case ImportKeys.Adaptor:
case ImportKeys.Sequence:
case ImportKeys.DbId:
case ImportKeys.Transcript:
// skip this key
break;
case ImportKeys.StartExon:
exonNode = node as ObjectKeyValueNode;
if (exonNode != null)
{
startExon = ImportExon.Parse(exonNode.Value, currentChromosome);
}
break;
case ImportKeys.EndExon:
exonNode = node as ObjectKeyValueNode;
if (exonNode != null)
{
endExon = ImportExon.Parse(exonNode.Value, currentChromosome);
}
break;
case ImportKeys.StableId:
proteinId = node.GetString();
break;
case ImportKeys.End:
end = node.GetInt32();
break;
case ImportKeys.Start:
start = node.GetInt32();
break;
case ImportKeys.Version:
proteinVersion = (byte)node.GetInt32();
break;
default:
throw new InvalidDataException($"Unknown key found: {node.Key}");
}
}
return(start, end, proteinId, proteinVersion, startExon, endExon);
}
/// <summary>
/// parses the relevant data from each transcript
/// </summary>
public static MutableTranscript Parse(ObjectValueNode objectValue, IChromosome chromosome, Source source)
{
// IDs
string transcriptId = null;
byte transcriptVersion = 1;
string proteinId = null;
byte proteinVersion = 0;
string ccdsId = null;
string refSeqId = null;
string geneId = null;
int hgncId = -1;
// gene
int geneStart = -1;
int geneEnd = -1;
var geneOnReverseStrand = false;
string geneSymbol = null;
var geneSymbolSource = GeneSymbolSource.Unknown;
// translation
int translationStart = -1;
int translationEnd = -1;
MutableExon translationStartExon = null;
MutableExon translationEndExon = null;
// predictions
string siftData = null;
string polyphenData = null;
var bioType = BioType.other;
IInterval[] microRnas = null;
MutableTranscriptRegion[] cdnaMaps = null;
IInterval[] introns = null;
string peptideSequence = null;
string translateableSequence = null;
var isCanonical = false;
int compDnaCodingStart = -1;
int compDnaCodingEnd = -1;
int start = -1;
int end = -1;
MutableExon[] exons = null;
var cdsStartNotFound = false;
var cdsEndNotFound = false;
int[] selenocysteinePositions = null;
IRnaEdit[] rnaEdits = null;
string bamEditStatus = null;
foreach (var node in objectValue.Values)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(node.Key))
{
throw new InvalidDataException($"Encountered an unknown key in the dumper transcript object: {node.Key}");
}
// handle each key
switch (node.Key)
{
case ImportKeys.CodingRegionEnd:
case ImportKeys.CodingRegionStart:
case ImportKeys.CreatedDate:
case ImportKeys.DbId:
case ImportKeys.Description:
case ImportKeys.DisplayXref:
case ImportKeys.ExternalDb:
case ImportKeys.ExternalDisplayName:
case ImportKeys.ExternalName:
case ImportKeys.ExternalStatus:
case ImportKeys.GenePhenotype:
case ImportKeys.GeneStableId:
case ImportKeys.ModifiedDate:
case ImportKeys.Protein:
case ImportKeys.Slice:
case ImportKeys.Source:
case ImportKeys.Strand:
case ImportKeys.SwissProt:
case ImportKeys.Trembl:
case ImportKeys.UniParc:
case ImportKeys.VepLazyLoaded:
// not used
break;
case ImportKeys.BamEditStatus:
bamEditStatus = node.GetString();
break;
case ImportKeys.Attributes:
(microRnas, rnaEdits, cdsStartNotFound, cdsEndNotFound) = Attribute.ParseList(node);
break;
case ImportKeys.Biotype:
bioType = TranscriptUtilities.GetBiotype(node);
break;
case ImportKeys.Ccds:
ccdsId = node.GetString();
break;
case ImportKeys.CdnaCodingEnd:
compDnaCodingEnd = node.GetInt32();
break;
case ImportKeys.CdnaCodingStart:
compDnaCodingStart = node.GetInt32();
break;
case ImportKeys.End:
end = node.GetInt32();
break;
case ImportKeys.GeneHgncId:
hgncId = node.GetHgncId();
break;
case ImportKeys.GeneSymbol:
case ImportKeys.GeneHgnc: // older key
geneSymbol = node.GetString();
break;
case ImportKeys.GeneSymbolSource:
geneSymbolSource = GeneSymbolSourceHelper.GetGeneSymbolSource(node.GetString());
break;
case ImportKeys.Gene:
(geneStart, geneEnd, geneId, geneOnReverseStrand) = ImportGene.Parse(node);
break;
case ImportKeys.IsCanonical:
isCanonical = node.GetBool();
break;
case ImportKeys.Refseq:
refSeqId = node.GetString();
break;
case ImportKeys.StableId:
transcriptId = node.GetString();
break;
case ImportKeys.Start:
start = node.GetInt32();
break;
case ImportKeys.TransExonArray:
exons = ImportExon.ParseList(node, chromosome);
break;
case ImportKeys.Translation:
(translationStart, translationEnd, proteinId, proteinVersion, translationStartExon, translationEndExon) = ImportTranslation.Parse(node, chromosome);
break;
case ImportKeys.VariationEffectFeatureCache:
(cdnaMaps, introns, peptideSequence, translateableSequence, siftData, polyphenData, selenocysteinePositions) = ImportVariantEffectFeatureCache.Parse(node);
break;
case ImportKeys.Version:
transcriptVersion = (byte)node.GetInt32();
break;
default:
throw new InvalidDataException($"Unknown key found: {node.Key}");
}
}
var fixedTranscript = AccessionUtilities.GetMaxVersion(transcriptId, transcriptVersion);
var fixedProtein = AccessionUtilities.GetMaxVersion(proteinId, proteinVersion);
var gene = new MutableGene(chromosome, geneStart, geneEnd, geneOnReverseStrand, geneSymbol,
geneSymbolSource, geneId, hgncId);
var codingRegion = new CodingRegion(GetCodingRegionStart(geneOnReverseStrand, translationStartExon, translationEndExon, translationStart, translationEnd),
GetCodingRegionEnd(geneOnReverseStrand, translationStartExon, translationEndExon, translationStart, translationEnd),
compDnaCodingStart, compDnaCodingEnd, 0);
int totalExonLength = GetTotalExonLength(exons);
int startExonPhase = translationStartExon?.Phase ?? int.MinValue;
return(new MutableTranscript(chromosome, start, end, fixedTranscript.Id, fixedTranscript.Version, ccdsId,
refSeqId, bioType, isCanonical, codingRegion, fixedProtein.Id, fixedProtein.Version,
peptideSequence, source, gene, exons, startExonPhase, totalExonLength, introns, cdnaMaps,
siftData, polyphenData, translateableSequence, microRnas, cdsStartNotFound, cdsEndNotFound,
selenocysteinePositions, rnaEdits, bamEditStatus));
}
private static bool ExonEquals(MutableExon x, MutableExon y)
{
return(x.Start == y.Start &&
x.End == y.End &&
x.Phase == y.Phase);
}