private static SupplementaryAnnotationHeader GetSaHeader(string saDir)
{
var saFiles = Directory.GetFiles(saDir, "*.nsa");
if (saFiles == null)
{
throw new UserErrorException($"Unable to find any supplementary annotation files in the following directory: {saDir}");
}
long intervalsPosition;
return(saFiles.Length > 0 ? SupplementaryAnnotationReader.GetHeader(saFiles.First(), out intervalsPosition) : null);
}
/// <summary>
/// checks the supplementary annotation directory to ensure that the directory only contains files for one
/// cache schema (8), one VEP version (79), and one genome assembly (GRCh37)
/// </summary>
public static void CheckDirectoryIntegrity(string saDir, List <DataSourceVersion> mainDataSourceVersions, out SupplementaryAnnotationDirectory saDirectory)
{
// sanity check: make sure the directory is set
if (string.IsNullOrEmpty(saDir))
{
saDirectory = null;
return;
}
var observedDataVersions = new HashSet <ushort>();
var refSeqsToPaths = new Dictionary <string, string>();
var observedGenomeAssemblies = new HashSet <GenomeAssembly>();
var earliestCreationTime = long.MaxValue;
var hasDataSourceVersions = false;
var foundDifferentDataSourceVersions = false;
var dataSourceVersions = new List <DataSourceVersion>();
// grab the header data from each cache file
foreach (var saPath in Directory.GetFiles(saDir, "*.nsa"))
{
long intervalsPosition;
var header = SupplementaryAnnotationReader.GetHeader(saPath, out intervalsPosition);
if (header == null)
{
continue;
}
observedDataVersions.Add(header.DataVersion);
refSeqsToPaths[header.ReferenceSequenceName] = saPath;
observedGenomeAssemblies.Add(header.GenomeAssembly);
if (header.CreationTimeTicks < earliestCreationTime)
{
earliestCreationTime = header.CreationTimeTicks;
}
if (hasDataSourceVersions)
{
if (!dataSourceVersions.SequenceEqual(header.DataSourceVersions))
{
foundDifferentDataSourceVersions = true;
}
}
else
{
dataSourceVersions = header.DataSourceVersions;
hasDataSourceVersions = true;
}
}
// sanity check: no references were found
if (refSeqsToPaths.Count == 0 || observedDataVersions.Count == 0)
{
throw new UserErrorException($"Unable to find any supplementary annotation files in the following directory: {saDir}");
}
// sanity check: more than one cache data version found
if (observedDataVersions.Count > 1)
{
throw new UserErrorException($"Found more than one cache data version represented in the following directory: {saDir}");
}
// sanity check: make sure all of the files have the same data source versions
if (foundDifferentDataSourceVersions)
{
throw new UserErrorException($"Found more than one set of data source versions represented in the following directory: {saDir}");
}
if (observedGenomeAssemblies.Count > 1)
{
throw new UserErrorException($"Found more than one set of Genome Assemblies represented in the following directory: {saDir}");
}
saDirectory = new SupplementaryAnnotationDirectory(observedDataVersions.First(), observedGenomeAssemblies.First());
mainDataSourceVersions.AddRange(dataSourceVersions);
}
