/// <summary>
/// parses the relevant data from each intron object
/// </summary>
private static void ParseReference(ObjectValue objectValue, DataStructures.VEP.Intron intron, ImportDataStore dataStore)
{
// loop over all of the key/value pairs in the intron object
foreach (AbstractData ad in objectValue)
{
// skip normal entries
if (!DumperUtilities.IsReference(ad))
{
continue;
}
// handle each key
switch (ad.Key)
{
case Transcript.SliceKey:
var referenceKeyValue = ad as ReferenceKeyValue;
if (referenceKeyValue != null)
{
intron.Slice = Slice.ParseReference(referenceKeyValue.Value, dataStore);
}
break;
default:
throw new GeneralException($"Found an unhandled reference in the intron object: {ad.Key}");
}
}
}
/// <summary>
/// parses the relevant data from each protein function prediction object
/// </summary>
public static void ParseReference(ObjectValue objectValue, DataStructures.VEP.ProteinFunctionPredictions cache, ImportDataStore dataStore)
{
// loop over all of the key/value pairs in the cache object
foreach (AbstractData ad in objectValue)
{
if (!DumperUtilities.IsReference(ad))
{
continue;
}
// handle each key
var referenceKeyValue = ad as ReferenceKeyValue;
if (referenceKeyValue == null)
{
continue;
}
switch (referenceKeyValue.Key)
{
case PolyPhenHumVarKey:
cache.PolyPhen = PolyPhen.ParseReference(referenceKeyValue.Value, dataStore);
break;
case SiftKey:
cache.Sift = Sift.ParseReference(referenceKeyValue.Value, dataStore);
break;
default:
throw new GeneralException(
$"Found an unhandled reference in the protein function prediction object: {ad.Key}");
}
}
}
/// <summary>
/// parses the relevant data from each PolyPhen object
/// </summary>
public static DataStructures.VEP.PolyPhen Parse(ObjectValue objectValue)
{
string matrix = null;
// loop over all of the key/value pairs in the PolyPhen object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper PolyPhen object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case AnalysisKey:
case IsMatrixCompressedKey:
case PeptideLengthKey:
case SubAnalysisKey:
case TranslationMD5Key:
break;
case MatrixKey:
matrix = DumperUtilities.GetString(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(new DataStructures.VEP.PolyPhen(matrix));
}
/// <summary>
/// parses the relevant data from each regulatory element
/// </summary>
public static void Parse(ObjectValue objectValue, int regulatoryFeatureIndex, ImportDataStore dataStore)
{
// Console.WriteLine("*** Parse {0} ***", regulatoryFeatureIndex + 1);
int start = -1;
int end = -1;
string stableId = null;
string type = null;
// loop over all of the key/value pairs in the transcript object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException(
$"Encountered an unknown key in the dumper regulatory element object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case BoundLengthsKey:
case CellTypeCountKey:
case CellTypesKey:
case DbIdKey:
case DisplayLabelKey:
case HasEvidenceKey:
case ProjectedKey:
case SetKey:
case Transcript.StrandKey:
case Transcript.SliceKey:
// not used
break;
case FeatureTypeKey:
type = DumperUtilities.GetString(ad);
break;
case Transcript.EndKey:
end = DumperUtilities.GetInt32(ad);
break;
case Transcript.StableIdKey:
stableId = DumperUtilities.GetString(ad);
break;
case Transcript.StartKey:
start = DumperUtilities.GetInt32(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
dataStore.RegulatoryFeatures.Add(new DataStructures.VEP.RegulatoryFeature(dataStore.CurrentReferenceIndex, start, end, stableId, type));
}
/// <summary>
/// returns a new exon given an ObjectValue
/// </summary>
public static DataStructures.VEP.Exon Parse(ObjectValue objectValue, ushort currentReferenceIndex)
{
bool onReverseStrand = false;
int end = -1;
byte?phase = null;
int start = -1;
string stableId = null;
// loop over all of the key/value pairs in the exon object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper mapper object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case Transcript.EndKey:
end = DumperUtilities.GetInt32(ad);
break;
case EndPhaseKey:
break;
case PhaseKey:
int phaseInt = DumperUtilities.GetInt32(ad);
if (phaseInt != -1)
{
phase = (byte)phaseInt;
}
break;
case Transcript.StableIdKey:
stableId = DumperUtilities.GetString(ad);
break;
case Transcript.StartKey:
start = DumperUtilities.GetInt32(ad);
break;
case Transcript.StrandKey:
onReverseStrand = TranscriptUtilities.GetStrand(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(new DataStructures.VEP.Exon(currentReferenceIndex, start, end, stableId, onReverseStrand, phase));
}
/// <summary>
/// parses the relevant data from each attribute
/// </summary>
private static SimpleInterval Parse(ObjectValue objectValue)
{
string key = null;
string value = null;
// loop over all of the key/value pairs in the gene object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException(
$"Encountered an unknown key in the dumper attribute object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case NameKey:
case DescriptionKey:
// not used
break;
case CodeKey:
key = DumperUtilities.GetString(ad);
break;
case ValueKey:
value = DumperUtilities.GetString(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
// sanity check: make sure this is a miRNA object
if (key != "miRNA" || value == null)
{
// Console.WriteLine("DEBUG: Found alternate attribute: {0}", key);
return(null);
}
var rangeMatch = RangeRegex.Match(value);
if (!rangeMatch.Success)
{
throw new GeneralException("Unable to convert the Attribute to a miRNA object. The value string failed the regex: " + value);
}
int start = int.Parse(rangeMatch.Groups[1].Value);
int end = int.Parse(rangeMatch.Groups[2].Value);
return(new SimpleInterval(start, end));
}
/// <summary>
/// parses the relevant data from each mapper pairs object
/// </summary>
public static List <DataStructures.VEP.MapperPair> ParseList(List <AbstractData> abstractDataList, ImportDataStore dataStore)
{
var mapperPairs = DumperUtilities.GetPopulatedList <DataStructures.VEP.MapperPair>(abstractDataList.Count);
// loop over all of the key/value pairs in the mapper pairs object
for (int mapperPairIndex = 0; mapperPairIndex < abstractDataList.Count; mapperPairIndex++)
{
var ad = abstractDataList[mapperPairIndex];
// skip references
if (DumperUtilities.IsReference(ad))
{
continue;
}
if (ad.DataType != DataType)
{
throw new GeneralException(
$"Expected a mapper pair data type, but found the following data type: [{ad.DataType}]");
}
var mapperPairNode = ad as ObjectValue;
if (mapperPairNode == null)
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectValue: [{ad.GetType()}]");
}
var newMapperPair = Parse(mapperPairNode, dataStore.CurrentReferenceIndex);
// DS.VEP.MapperPair oldMapperPair;
// if (dataStore.MapperPairs.TryGetValue(newMapperPair, out oldMapperPair))
//{
// mapperPairs[mapperPairIndex] = oldMapperPair;
//}
// else
//{
mapperPairs[mapperPairIndex] = newMapperPair;
// dataStore.MapperPairs[newMapperPair] = newMapperPair;
//}
}
return(mapperPairs);
}
/// <summary>
/// places a reference to already existing exons into the array of exons
/// </summary>
public static void ParseListReference(List <AbstractData> abstractDataList, DataStructures.VEP.Exon[] exons, ImportDataStore dataStore)
{
// loop over all of the exons
for (int exonIndex = 0; exonIndex < abstractDataList.Count; exonIndex++)
{
var exonNode = abstractDataList[exonIndex];
// skip normal exons
if (!DumperUtilities.IsReference(exonNode))
{
continue;
}
var referenceStringValue = exonNode as ReferenceStringValue;
if (referenceStringValue != null)
{
exons[exonIndex] = ParseReference(referenceStringValue.Value, dataStore);
}
}
}
/// <summary>
/// returns an array of miRNAs given a list of ObjectValues (AbstractData)
/// </summary>
public static SimpleInterval[] ParseList(List <AbstractData> abstractDataList)
{
var microRnas = new List <SimpleInterval>();
foreach (var ad in abstractDataList)
{
// skip references
if (DumperUtilities.IsReference(ad))
{
continue;
}
var objectValue = ad as ObjectValue;
if (objectValue != null)
{
var newMicroRna = Parse(objectValue);
if (newMicroRna != null)
{
// MicroRna oldMicroRna;
// if (dataStore.MicroRnas.TryGetValue(newMicroRna, out oldMicroRna))
//{
// microRnas.Add(oldMicroRna);
//}
// else
//{
microRnas.Add(newMicroRna);
// dataStore.MicroRnas[newMicroRna] = newMicroRna;
//}
}
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectValue: [{ad.GetType()}]");
}
}
return(microRnas.Count == 0 ? null : microRnas.ToArray());
}
/// <summary>
/// points to a translation that has already been created
/// </summary>
public static void ParseReference(ObjectValue objectValue, DataStructures.VEP.Translation translation, ImportDataStore dataStore)
{
// loop over all of the key/value pairs in the translation object
foreach (AbstractData ad in objectValue)
{
if (!DumperUtilities.IsReference(ad))
{
continue;
}
// handle each key
var referenceKeyValue = ad as ReferenceKeyValue;
if (referenceKeyValue == null)
{
continue;
}
switch (referenceKeyValue.Key)
{
case AdaptorKey:
// skip this key
break;
case EndExonKey:
translation.EndExon = Exon.ParseReference(referenceKeyValue.Value, dataStore);
break;
case StartExonKey:
translation.StartExon = Exon.ParseReference(referenceKeyValue.Value, dataStore);
break;
case TranscriptKey:
translation.Transcript = Transcript.ParseReference(referenceKeyValue.Value, dataStore);
break;
default:
throw new GeneralException(
$"Found an unhandled reference in the translation object: {ad.Key}");
}
}
}
/// <summary>
/// parses the relevant data from each mapper pairs object
/// </summary>
public static void ParseListReference(List <AbstractData> abstractDataList, List <DataStructures.VEP.MapperPair> mapperPairs, ImportDataStore dataStore)
{
// loop over all of the key/value pairs in the mapper pairs object
for (int mapperPairIndex = 0; mapperPairIndex < abstractDataList.Count; mapperPairIndex++)
{
var mapperNode = abstractDataList[mapperPairIndex];
// skip normal mapper pairs
if (!DumperUtilities.IsReference(mapperNode))
{
continue;
}
var referenceStringValue = mapperNode as ReferenceStringValue;
if (referenceStringValue != null)
{
var mapperPair = ParseReference(referenceStringValue.Value, dataStore);
mapperPairs[mapperPairIndex] = mapperPair;
}
}
}
/// <summary>
/// parses the relevant data from each pair genomic object
/// </summary>
public static DataStructures.VEP.PairGenomic Parse(ObjectValue objectValue, ImportDataStore dataStore)
{
var pairGenomic = new DataStructures.VEP.PairGenomic();
// loop over all of the key/value pairs in the pair genomic object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the pair genomic object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case GenomicKey:
var genomicNode = ad as ListObjectKeyValue;
if (genomicNode != null)
{
pairGenomic.Genomic = MapperPair.ParseList(genomicNode.Values, dataStore);
}
else if (DumperUtilities.IsUndefined(ad))
{
pairGenomic.Genomic = null;
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(pairGenomic);
}
/// <summary>
/// returns an array of exons given a list of ObjectValues (AbstractData)
/// </summary>
public static DataStructures.VEP.Exon[] ParseList(List <AbstractData> abstractDataList, ImportDataStore dataStore)
{
var exons = new DataStructures.VEP.Exon[abstractDataList.Count];
// loop over all of the exons
for (int exonIndex = 0; exonIndex < abstractDataList.Count; exonIndex++)
{
// skip references
if (DumperUtilities.IsReference(abstractDataList[exonIndex]))
{
continue;
}
var objectValue = abstractDataList[exonIndex] as ObjectValue;
if (objectValue != null)
{
var newExon = Parse(objectValue, dataStore.CurrentReferenceIndex);
// DS.VEP.Exon oldExon;
// if (dataStore.Exons.TryGetValue(newExon, out oldExon))
//{
// exons[exonIndex] = oldExon;
//}
// else
//{
exons[exonIndex] = newExon;
// dataStore.Exons[newExon] = newExon;
//}
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectValue: [{abstractDataList[exonIndex].GetType()}]");
}
}
return(exons);
}
/// <summary>
/// parses the relevant data from each mapper unit object
/// </summary>
public static DataStructures.VEP.MapperUnit Parse(ObjectValue objectValue, ushort currentReferenceIndex)
{
var id = MapperUnitType.Unknown;
int end = -1;
int start = -1;
// loop over all of the key/value pairs in the mapper unit object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the mapper unit object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case Transcript.EndKey:
end = DumperUtilities.GetInt32(ad);
break;
case IdKey:
id = TranscriptUtilities.GetMapperUnitType(ad);
break;
case Transcript.StartKey:
start = DumperUtilities.GetInt32(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(new DataStructures.VEP.MapperUnit(currentReferenceIndex, start, end, id));
}
/// <summary>
/// returns a new exon given an ObjectValue
/// </summary>
private static DataStructures.VEP.Intron Parse(ObjectValue objectValue, ImportDataStore dataStore)
{
var intron = new DataStructures.VEP.Intron();
// loop over all of the key/value pairs in the intron object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper mapper object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case Transcript.EndKey:
intron.End = DumperUtilities.GetInt32(ad);
break;
case Transcript.SliceKey:
var sliceNode = ad as ObjectKeyValue;
if (sliceNode != null)
{
var newSlice = Slice.Parse(sliceNode.Value, dataStore.CurrentReferenceIndex);
// DS.VEP.Slice oldSlice;
// if (dataStore.Slices.TryGetValue(newSlice, out oldSlice))
//{
// intron.Slice = oldSlice;
//}
// else
//{
intron.Slice = newSlice;
// dataStore.Slices[newSlice] = newSlice;
//}
}
else if (DumperUtilities.IsReference(ad))
{
// skip references until the second pass
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue or ReferenceKeyValue: [{ad.GetType()}]");
}
break;
case Transcript.StartKey:
intron.Start = DumperUtilities.GetInt32(ad);
break;
case Transcript.StrandKey:
TranscriptUtilities.GetStrand(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(intron);
}
/// <summary>
/// parses the relevant data from each exon coordinate mapper object
/// </summary>
public static DataStructures.VEP.Mapper Parse(ObjectValue objectValue, ImportDataStore dataStore)
{
var mapper = new DataStructures.VEP.Mapper();
// loop over all of the key/value pairs in the exon coordinate mapper object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper mapper object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case FromCoordSystemKey:
if (!DumperUtilities.IsUndefined(ad))
{
throw new GeneralException("Found an unexpected value in FromCoordSystemKey");
}
break;
case FromNameKey:
mapper.FromType = DumperUtilities.GetString(ad);
break;
case IsSortedKey:
mapper.IsSorted = DumperUtilities.GetBool(ad);
break;
case PairCodingDnaKey:
var pairCodingDnaNode = ad as ObjectKeyValue;
if (pairCodingDnaNode != null)
{
mapper.PairCodingDna = PairCodingDna.Parse(pairCodingDnaNode.Value, dataStore);
}
else if (DumperUtilities.IsUndefined(ad))
{
mapper.PairCodingDna = null;
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
case PairCountKey:
mapper.PairCount = DumperUtilities.GetInt32(ad);
break;
case PairGenomicKey:
var pairGenomicNode = ad as ObjectKeyValue;
if (pairGenomicNode != null)
{
mapper.PairGenomic = PairGenomic.Parse(pairGenomicNode.Value, dataStore);
}
else if (DumperUtilities.IsUndefined(ad))
{
mapper.PairGenomic = null;
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
case ToCoordSystemKey:
if (!DumperUtilities.IsUndefined(ad))
{
throw new GeneralException("Found an unexpected value in ToCoordSystemKey");
}
break;
case ToNameKey:
mapper.ToType = DumperUtilities.GetString(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(mapper);
}
/// <summary>
/// parses the relevant data from each variant effect feature cache
/// </summary>
public static DataStructures.VEP.VariantEffectFeatureCache Parse(ObjectValue objectValue, ImportDataStore dataStore)
{
var cache = new DataStructures.VEP.VariantEffectFeatureCache();
// loop over all of the key/value pairs in the cache object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException(
$"Encountered an unknown key in the dumper variant effect feature cache object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case SelenocysteinesKey:
case ThreePrimeUtrKey:
case SeqEditsKey:
case CodonTableKey:
case ProteinFeaturesKey:
// not used
break;
case IntronsKey:
var intronsList = ad as ListObjectKeyValue;
if (intronsList != null)
{
cache.Introns = Intron.ParseList(intronsList.Values, dataStore);
}
else if (DumperUtilities.IsUndefined(ad))
{
cache.Introns = null;
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into a ListObjectKeyValue: [{ad.GetType()}]");
}
break;
case MapperKey:
var mapperNode = ad as ObjectKeyValue;
if (mapperNode != null)
{
cache.Mapper = TranscriptMapper.Parse(mapperNode.Value, dataStore);
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
case PeptideKey:
cache.Peptide = DumperUtilities.GetString(ad);
break;
case ProteinFunctionPredictionsKey:
var predictionsNode = ad as ObjectKeyValue;
if (predictionsNode != null)
{
cache.ProteinFunctionPredictions = ProteinFunctionPredictions.Parse(predictionsNode.Value);
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
case SortedExonsKey:
var exonsList = ad as ListObjectKeyValue;
if (exonsList != null)
{
cache.Exons = Exon.ParseList(exonsList.Values, dataStore);
}
else
{
throw new GeneralException($"Could not transform the AbstractData object into a ListObjectKeyValue: [{ad.GetType()}]");
}
break;
case TranslateableSeqKey:
cache.TranslateableSeq = DumperUtilities.GetString(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(cache);
}
/// <summary>
/// parses the relevant data from each transcript
/// </summary>
public static void Parse(ObjectValue objectValue, int transcriptIndex, ImportDataStore dataStore)
{
// Console.WriteLine("*** Parse {0} ***", transcriptIndex + 1);
var bioType = BioType.Unknown;
var geneSymbolSource = GeneSymbolSource.Unknown; // HGNC
SimpleInterval[] microRnas = null;
DataStructures.VEP.Exon[] transExons = null;
DataStructures.VEP.Gene gene = null;
DataStructures.VEP.Translation translation = null;
DataStructures.VEP.VariantEffectFeatureCache variantEffectCache = null;
DataStructures.VEP.Slice slice = null;
bool onReverseStrand = false;
bool isCanonical = false;
int compDnaCodingStart = -1;
int compDnaCodingEnd = -1;
int start = -1;
int end = -1;
byte version = 1;
string ccdsId = null;
string databaseId = null;
string proteinId = null;
string refSeqId = null;
string geneStableId = null;
string stableId = null;
string geneSymbol = null; // DDX11L1
int hgncId = -1; // 37102
// loop over all of the key/value pairs in the transcript object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper transcript object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case CodingRegionStartKey:
case CodingRegionEndKey:
case CreatedDateKey:
case DescriptionKey:
case DisplayXrefKey:
case ExternalDbKey:
case ExternalDisplayNameKey:
case ExternalNameKey:
case ExternalStatusKey:
case GenePhenotypeKey:
case ModifiedDateKey:
case SourceKey:
case SwissProtKey:
case TremblKey:
case UniParcKey:
// not used
break;
case AttributesKey:
var attributesList = ad as ListObjectKeyValue;
if (attributesList != null)
{
microRnas = Attribute.ParseList(attributesList.Values);
}
break;
case BiotypeKey:
bioType = TranscriptUtilities.GetBiotype(ad);
break;
case CcdsKey:
ccdsId = DumperUtilities.GetString(ad);
if (ccdsId == "-" || ccdsId == "")
{
ccdsId = null;
}
break;
case CdnaCodingEndKey:
compDnaCodingEnd = DumperUtilities.GetInt32(ad);
break;
case CdnaCodingStartKey:
compDnaCodingStart = DumperUtilities.GetInt32(ad);
break;
case DbIdKey:
databaseId = DumperUtilities.GetString(ad);
if (databaseId == "-" || databaseId == "")
{
databaseId = null;
}
break;
case EndKey:
end = DumperUtilities.GetInt32(ad);
break;
case GeneHgncIdKey:
var hgnc = DumperUtilities.GetString(ad);
if (hgnc != null && hgnc.StartsWith("HGNC:"))
{
hgnc = hgnc.Substring(5);
}
if (hgnc == "-" || hgnc == "")
{
hgnc = null;
}
if (hgnc != null)
{
hgncId = int.Parse(hgnc);
}
break;
case GeneSymbolKey:
case GeneHgncKey: // older key
geneSymbol = DumperUtilities.GetString(ad);
if (geneSymbol == "-" || geneSymbol == "")
{
geneSymbol = null;
}
break;
case GeneSymbolSourceKey:
geneSymbolSource = TranscriptUtilities.GetGeneSymbolSource(ad);
break;
case GeneKey:
var geneNode = ad as ObjectKeyValue;
if (geneNode != null)
{
gene = Gene.Parse(geneNode.Value, dataStore.CurrentReferenceIndex);
}
break;
case GeneStableIdKey:
geneStableId = DumperUtilities.GetString(ad);
if (geneStableId == "-" || geneStableId == "")
{
geneStableId = null;
}
break;
case IsCanonicalKey:
isCanonical = DumperUtilities.GetBool(ad);
break;
case ProteinKey:
proteinId = DumperUtilities.GetString(ad);
if (proteinId == "-" || proteinId == "")
{
proteinId = null;
}
break;
case RefseqKey:
refSeqId = DumperUtilities.GetString(ad);
if (refSeqId == "-" || refSeqId == "")
{
refSeqId = null;
}
break;
case SliceKey:
var sliceNode = ad as ObjectKeyValue;
if (sliceNode != null)
{
slice = Slice.Parse(sliceNode.Value, dataStore.CurrentReferenceIndex);
}
break;
case StableIdKey:
stableId = DumperUtilities.GetString(ad);
if (stableId == "-" || stableId == "")
{
stableId = null;
}
break;
case StartKey:
start = DumperUtilities.GetInt32(ad);
break;
case StrandKey:
onReverseStrand = TranscriptUtilities.GetStrand(ad);
break;
case TransExonArrayKey:
var exonsList = ad as ListObjectKeyValue;
if (exonsList != null)
{
transExons = Exon.ParseList(exonsList.Values, dataStore);
}
else
{
throw new GeneralException($"Could not transform the AbstractData object into a ListObjectKeyValue: [{ad.GetType()}]");
}
break;
case TranslationKey:
var translationNode = ad as ObjectKeyValue;
if (translationNode != null)
{
translation = Translation.Parse(translationNode.Value, dataStore);
}
else if (DumperUtilities.IsUndefined(ad))
{
translation = null;
}
else
{
throw new GeneralException($"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
case VariationEffectFeatureCacheKey:
var cacheNode = ad as ObjectKeyValue;
if (cacheNode == null)
{
throw new GeneralException($"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
variantEffectCache = VariantEffectFeatureCache.Parse(cacheNode.Value, dataStore);
break;
case VersionKey:
version = (byte)DumperUtilities.GetInt32(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
dataStore.Transcripts.Add(new DataStructures.VEP.Transcript(bioType, transExons, gene, translation, variantEffectCache, slice,
onReverseStrand, isCanonical, compDnaCodingStart, compDnaCodingEnd, dataStore.CurrentReferenceIndex, start, end,
ccdsId, databaseId, proteinId, refSeqId, geneStableId, stableId, geneSymbol, geneSymbolSource, hgncId, version,
microRnas));
}
/// <summary>
/// parses the relevant data from each transcript
/// </summary>
public static void ParseReferences(ObjectValue objectValue, int transcriptIndex, ImportDataStore dataStore)
{
// Console.WriteLine("*** ParseReferences {0} / {1} ***", transcriptIndex + 1, _tempTranscripts.Count);
var transcript = dataStore.Transcripts[transcriptIndex];
// loop over all of the key/value pairs in the transcript object
foreach (AbstractData ad in objectValue)
{
// skip undefined keys
if (DumperUtilities.IsUndefined(ad))
{
continue;
}
// handle each key
ReferenceKeyValue referenceKeyValue;
// references found in:
// 'transcript' -> '_variation_effect_feature_cache' -> 'introns' -> 'slice' has references
// 'transcript' -> 'gene' has references
// 'transcript' -> 'slice' has references
// 'transcript' -> '_trans_exon_array' -> [] has references
// 'transcript' -> 'translation'-> 'end_exon' has references
// 'transcript' -> 'translation'-> 'start_exon' has references
// 'transcript' -> 'translation'-> 'transcript' has references
switch (ad.Key)
{
case GeneKey:
// works well
if (DumperUtilities.IsReference(ad))
{
referenceKeyValue = ad as ReferenceKeyValue;
if (referenceKeyValue != null)
{
transcript.Gene = Gene.ParseReference(referenceKeyValue.Value, dataStore);
}
}
break;
case SliceKey:
if (DumperUtilities.IsReference(ad))
{
referenceKeyValue = ad as ReferenceKeyValue;
if (referenceKeyValue != null)
{
transcript.Slice = Slice.ParseReference(referenceKeyValue.Value, dataStore);
}
}
break;
case TransExonArrayKey:
var exonsList = ad as ListObjectKeyValue;
if (exonsList != null)
{
Exon.ParseListReference(exonsList.Values, transcript.TransExons, dataStore);
}
break;
case TranslationKey:
var translationNode = ad as ObjectKeyValue;
if (translationNode != null)
{
Translation.ParseReference(translationNode.Value, transcript.Translation, dataStore);
}
break;
case VariationEffectFeatureCacheKey:
var cacheNode = ad as ObjectKeyValue;
if (cacheNode != null)
{
VariantEffectFeatureCache.ParseReference(cacheNode.Value, transcript.VariantEffectCache, dataStore);
}
break;
}
}
}
/// <summary>
/// parses the relevant data from each slice
/// </summary>
public static DataStructures.VEP.Slice Parse(ObjectValue objectValue, ushort currentReferenceIndex)
{
DataStructures.VEP.CoordSystem coordinateSystem = null;
bool isCircular = false;
bool isTopLevel = false;
bool onReverseStrand = false;
int start = -1;
int end = -1;
int sequenceRegionLen = -1;
string sequenceRegionName = null;
// loop over all of the key/value pairs in the gene object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper slice object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case CoordSystemKey:
var coordSystemNode = ad as ObjectKeyValue;
if (coordSystemNode != null)
{
coordinateSystem = CoordSystem.Parse(coordSystemNode.Value);
}
break;
case CircularKey:
isCircular = DumperUtilities.GetBool(ad);
break;
case Transcript.EndKey:
end = DumperUtilities.GetInt32(ad);
break;
case SequenceRegionLenKey:
sequenceRegionLen = DumperUtilities.GetInt32(ad);
break;
case SequenceRegionNameKey:
sequenceRegionName = DumperUtilities.GetString(ad);
break;
case Transcript.StartKey:
start = DumperUtilities.GetInt32(ad);
break;
case Transcript.StrandKey:
onReverseStrand = TranscriptUtilities.GetStrand(ad);
break;
case TopLevelSliceKey:
isTopLevel = DumperUtilities.GetBool(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(new DataStructures.VEP.Slice(currentReferenceIndex, start, end, onReverseStrand, isCircular, isTopLevel, coordinateSystem, sequenceRegionLen, sequenceRegionName));
}
/// <summary>
/// parses the relevant data from each translation object
/// </summary>
public static DataStructures.VEP.Translation Parse(ObjectValue objectValue, ImportDataStore dataStore)
{
var translation = new DataStructures.VEP.Translation();
// loop over all of the key/value pairs in the translation object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper mapper object: {ad.Key}");
}
// handle each key
ObjectKeyValue exonNode;
switch (ad.Key)
{
case AdaptorKey:
case SequenceKey:
case Transcript.DbIdKey:
case Transcript.StableIdKey:
// skip this key
break;
case EndExonKey:
exonNode = ad as ObjectKeyValue;
if (exonNode != null)
{
var newExon = Exon.Parse(exonNode.Value, dataStore.CurrentReferenceIndex);
translation.EndExon = newExon;
}
break;
case StartExonKey:
exonNode = ad as ObjectKeyValue;
if (exonNode != null)
{
var newExon = Exon.Parse(exonNode.Value, dataStore.CurrentReferenceIndex);
translation.StartExon = newExon;
}
break;
case TranscriptKey:
// parse this during the references
if (!DumperUtilities.IsReference(ad))
{
throw new GeneralException("Found a Translation->Transcript entry that wasn't a reference.");
}
break;
case Transcript.EndKey:
translation.End = DumperUtilities.GetInt32(ad);
break;
case Transcript.StartKey:
translation.Start = DumperUtilities.GetInt32(ad);
break;
case Transcript.VersionKey:
translation.Version = (byte)DumperUtilities.GetInt32(ad);
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(translation);
}
/// <summary>
/// parses the relevant data from each protein function predictions object
/// </summary>
public static DataStructures.VEP.ProteinFunctionPredictions Parse(ObjectValue objectValue)
{
var predictions = new DataStructures.VEP.ProteinFunctionPredictions();
// loop over all of the key/value pairs in the protein function predictions object
foreach (AbstractData ad in objectValue)
{
// sanity check: make sure we know about the keys are used for
if (!KnownKeys.Contains(ad.Key))
{
throw new GeneralException($"Encountered an unknown key in the dumper mapper object: {ad.Key}");
}
// handle each key
switch (ad.Key)
{
case PolyPhenHumDivKey:
// not used by default
break;
case PolyPhenKey:
if (DumperUtilities.IsUndefined(ad))
{
// do nothing
}
else
{
throw new GeneralException($"Could not handle the PolyPhen key: [{ad.GetType()}]");
}
break;
case PolyPhenHumVarKey:
// used by default
var polyPhenHumVarNode = ad as ObjectKeyValue;
if (polyPhenHumVarNode != null)
{
predictions.PolyPhen = PolyPhen.Parse(polyPhenHumVarNode.Value);
}
else if (DumperUtilities.IsUndefined(ad))
{
predictions.PolyPhen = null;
}
else if (DumperUtilities.IsReference(ad))
{
// skip references for now
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
case SiftKey:
var siftNode = ad as ObjectKeyValue;
if (siftNode != null)
{
predictions.Sift = Sift.Parse(siftNode.Value);
}
else if (DumperUtilities.IsUndefined(ad))
{
predictions.Sift = null;
}
else if (DumperUtilities.IsReference(ad))
{
// skip references for now
}
else
{
throw new GeneralException(
$"Could not transform the AbstractData object into an ObjectKeyValue: [{ad.GetType()}]");
}
break;
default:
throw new GeneralException($"Unknown key found: {ad.Key}");
}
}
return(predictions);
}
