private static (Stream jsonStream, Stream jasixStream) GetNirvanaJsonStreamWithoutGenes(int chromNumber)
{
var jsonStream = new MemoryStream();
var jasixStream = new MemoryStream();
using (var bgZipStream = new BlockGZipStream(jsonStream, CompressionMode.Compress, true))
using (var writer = new BgzipTextWriter(bgZipStream))
using (var jasixIndex = new JasixIndex())
{
writer.Write(NirvanaHeader);
writer.Flush();
jasixIndex.BeginSection(JasixCommons.PositionsSectionTag, writer.Position);
for (int i = 100 * chromNumber; i < 123 * chromNumber; i++)
{
writer.WriteLine($"{{\"chromosome\":\"chr{chromNumber}\",\"position\":{i}}},");
if (i % 50 == 0)
{
writer.Flush(); //creating another block
}
}
writer.WriteLine($"{{\"chromosome\":\"chr{chromNumber}\",\"position\":{100 *chromNumber +25}}}");
writer.Flush();
jasixIndex.EndSection(JasixCommons.PositionsSectionTag, writer.Position);
writer.Write(NirvanaFooter);
jasixIndex.Write(jasixStream);
}
jsonStream.Position = 0;
jasixStream.Position = 0;
return(jsonStream, jasixStream);
}
public void CreateIndex()
{
var searchTag = $"\"{SectionToIndex}\":[";
const string headerTag = "{\"header\":";
var index = new JasixIndex();
string line;
//skipping lines before the sectionToIndex arrives
while ((line = _reader.ReadLine()) != null)
{
if (line.StartsWith(headerTag))
{
index.HeaderLine = ExtractHeader(line);
}
if (line.EndsWith(searchTag))
{
break;
}
}
// we need the location before accessing the line
var fileLoc = _reader.Position;
string previousChr = "";
int previousPos = 0;
while ((line = _reader.ReadLine()) != null)
{
if (line.StartsWith("]"))
{
break;
}
line = line.TrimEnd(',');
var chrPos = GetChromPosition(line);
CheckFileSorted(chrPos.chr, chrPos.position, previousChr, previousPos);
index.Add(chrPos.chr, chrPos.position, chrPos.end, fileLoc);
fileLoc = _reader.Position;
previousChr = chrPos.chr;
previousPos = chrPos.position;
}
index.Write(_writeStream);
Console.WriteLine();
var peakMemoryUsageBytes = MemoryUtilities.GetPeakMemoryUsage();
var wallTimeSpan = _benchmark.GetElapsedTime();
Console.WriteLine();
if (peakMemoryUsageBytes > 0)
{
Console.WriteLine("Peak memory usage: {0}", MemoryUtilities.ToHumanReadable(peakMemoryUsageBytes));
}
Console.WriteLine("Time: {0}", Benchmark.ToHumanReadable(wallTimeSpan));
}
public void Write_and_read_back()
{
var index = new JasixIndex();
index.Add("chr1", 100, 101, 100000, "1");
index.Add("chr1", 105, 109, 100050, "1");
index.Add("chr1", 150, 1000, 100075, "1");//large variant
index.Add("chr1", 160, 166, 100100, "1");
index.Add("chr2", 100, 100, 100150, "2");
index.Add("chr2", 102, 105, 100200, "2");
var writeStream = new MemoryStream();
using (writeStream)
{
index.Write(writeStream);
}
var readStream = new MemoryStream(writeStream.ToArray());
readStream.Seek(0, SeekOrigin.Begin);
JasixIndex readBackIndex;
using (readStream)
{
readBackIndex = new JasixIndex(readStream);
}
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 100, 102));
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 103, 104));
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 120, 124));
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 158, 160));
Assert.Equal(100150, readBackIndex.GetFirstVariantPosition("chr2", 103, 105));
//checking large variants
Assert.Null(readBackIndex.LargeVariantPositions("chr1", 100, 149));
var largeVariants = readBackIndex.LargeVariantPositions("chr1", 100, 201);
Assert.NotNull(largeVariants);
Assert.Single(largeVariants);
Assert.Equal(100075, largeVariants[0]);
}
public void IndexWriteRead()
{
var index = new JasixIndex();
index.Add("chr1", 100, 101, 100000);
index.Add("chr1", 105, 109, 100050);
index.Add("chr1", 150, 1000, 100075);//large variant
index.Add("chr1", 160, 166, 100100);
index.Add("chr2", 100, 100, 100150);
index.Add("chr2", 102, 105, 100200);
var tempFile = Path.Combine(Path.GetTempPath(), Path.GetRandomFileName());
using (var writer = FileUtilities.GetCreateStream(tempFile))
{
index.Write(writer);
}
JasixIndex readBackIndex;
using (var stream = FileUtilities.GetReadStream(tempFile))
{
readBackIndex = new JasixIndex(stream);
}
File.Delete(tempFile);
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 100, 102));
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 103, 104));
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 120, 124));
Assert.Equal(100000, readBackIndex.GetFirstVariantPosition("chr1", 158, 160));
Assert.Equal(100150, readBackIndex.GetFirstVariantPosition("chr2", 103, 105));
//checking large variants
Assert.Null(readBackIndex.LargeVariantPositions("chr1", 100, 149));
var largeVariants = readBackIndex.LargeVariantPositions("chr1", 100, 201);
Assert.NotNull(largeVariants);
Assert.Single(largeVariants);
Assert.Equal(100075, largeVariants[0]);
}
public void CreateIndex()
{
var index = new JasixIndex();
IndexHeader(index);
string lastLine = IndexPositions(index);
IndexGenes(lastLine, index);
index.Write(_writeStream);
Console.WriteLine();
long peakMemoryUsageBytes = MemoryUtilities.GetPeakMemoryUsage();
var wallTimeSpan = _benchmark.GetElapsedTime();
Console.WriteLine();
if (peakMemoryUsageBytes > 0)
{
Console.WriteLine("Peak memory usage: {0}", MemoryUtilities.ToHumanReadable(peakMemoryUsageBytes));
}
Console.WriteLine("Time: {0}", Benchmark.ToHumanReadable(wallTimeSpan));
}
public void Dispose()
{
_jasixIndex.Write(_indexStream);
_indexStream.Flush();
_indexStream.Dispose();
}
public void Flush()
{
_jasixIndex.Write(_indexStream);
_indexStream.Flush();
}