/// <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));
}