private string IndexPositions(JasixIndex index)
{
// we need the location before accessing the line
long linePosition = _reader.Position;
index.BeginSection(JasixCommons.PositionsSectionTag, linePosition);
Console.WriteLine($"section:{JasixCommons.PositionsSectionTag} starts at {linePosition}");
var previousChr = "";
var previousPos = 0;
string line;
while ((line = _reader.ReadLine()) != null)
{
if (line.OptimizedStartsWith(']'))
{
index.EndSection(JasixCommons.PositionsSectionTag, linePosition);
Console.WriteLine($"section:{JasixCommons.PositionsSectionTag} ends at {linePosition}");
break;
}
line = line.TrimEnd(',');
(string chr, int position, int end) = GetChromPosition(line);
CheckSorting(chr, position, previousChr, previousPos);
index.Add(chr, position, end, linePosition);
linePosition = _reader.Position;
previousChr = chr;
previousPos = position;
}
return(line);
}
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 GetFirstVariantPosition_multi_chrom_index()
{
var index = new JasixIndex();
index.Add("chr1", 100, 101, 100000);
index.Add("chr1", 105, 109, 100050);
index.Add("chr1", 160, 166, 100100);
index.Add("chr2", 100, 100, 100150);
index.Add("chr2", 102, 105, 100200);
index.Flush();
var chrPos = Utilities.ParseQuery("chr1");
Assert.Equal(100000, index.GetFirstVariantPosition(chrPos.Item1, chrPos.Item2, chrPos.Item3));
chrPos = Utilities.ParseQuery("chr2");
Assert.Equal(100150, index.GetFirstVariantPosition(chrPos.Item1, chrPos.Item2, chrPos.Item3));
}
public void Add_fill_node_and_start_another()
{
var index = new JasixIndex();
//creating two nodes each containing 50 entries
for (var i = 0; i < 2 * JasixCommons.PreferredNodeCount; i++)
{
index.Add("chr1", 100 + i, 101 + i, 100_000 + i);
}
index.Add("chr1", 160 + 2 * JasixCommons.PreferredNodeCount, 166 + 2 * JasixCommons.PreferredNodeCount, 200_100);
index.Add("chr2", 100, 100, 200_150);
index.Add("chr2", 102, 105, 200_200);
index.Flush();
Assert.Equal(100_000, index.GetFirstVariantPosition("chr1", 100, 102));
Assert.Equal(100_000 + JasixCommons.PreferredNodeCount, index.GetFirstVariantPosition("chr1", 2 * JasixCommons.PreferredNodeCount + 55, 2 * JasixCommons.PreferredNodeCount + 55));
Assert.Equal(-1, index.GetFirstVariantPosition("chr1", 2 * JasixCommons.PreferredNodeCount + 120, 2 * JasixCommons.PreferredNodeCount + 124));
Assert.Equal(200_100, index.GetFirstVariantPosition("chr1", 2 * JasixCommons.PreferredNodeCount + 158, 2 * JasixCommons.PreferredNodeCount + 160));
Assert.Equal(200_150, index.GetFirstVariantPosition("chr2", 103, 105));
}
public void Combination_of_large_and_small_variants()
{
var index = new JasixIndex();
//query range 10,000- 10,020
index.Add("chr1", 8_000, 9_900, 90_000); //SV not overlapping the query
index.Add("chr1", 9_000, 10_005, 90_100); // partially overlapping
index.Add("chr1", 9_500, 10_050, 90_200); //completely overlapping
index.Add("chr1", 10_000, 10_001, 100_000);
index.Add("chr1", 10_004, 10_006, 100_100);
index.Add("chr1", 10_009, 10_550, 100_200);//SV starting from the middle of the range
index.Add("chr1", 10_008, 10_010, 100_300);
index.Add("chr1", 10_011, 10_020, 100_400);
index.Add("chr1", 10_039, 10_550, 100_200);//SV past the range
index.Flush();
var firstSmallVarLocation = index.GetFirstVariantPosition("chr1", 10_000, 10_020);
var largeVariantLocations = index.LargeVariantPositions("chr1", 10_000, 10_020);
Assert.Equal(90_000, firstSmallVarLocation);
Assert.True(largeVariantLocations.SequenceEqual(new List <long> {
90_100, 90_200, 100_200
}));
public void Add(IPosition position, long fileLocation)
{
string chromName = position.Chromosome.EnsemblName;
int start = position.Start;
int? end = position.InfoData?.End;
if (chromName == _lastChromName && start < _lastPosition)
{
throw new UserErrorException($"The Json file is not sorted at {position.Chromosome.UcscName}: {start}");
}
_lastPosition = start;
_lastChromName = chromName;
if (end == null)
{
string[] altAlleles = position.AltAlleles;
int altAlleleOffset = altAlleles != null && altAlleles.All(Utilities.IsNucleotideAllele) && altAlleles.Any(x => x.Length > 1) ? 1 : 0;
end = Math.Max(position.RefAllele.Length - 1, altAlleleOffset) + start;
}
_jasixIndex.Add(position.Chromosome.EnsemblName, start, end.Value, fileLocation, position.Chromosome.UcscName);
}
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 FindLargeVaritants_method_does_not_return_small_variants()
{
var index = new JasixIndex();
index.Add("chr1", 100, 101, 100_000);
index.Add("chr1", 105, 109, 100_050);
index.Add("chr1", 160, 166, 100_100);
index.Add("chr1", 200, 1000, 100_075);//large variant
index.Add("chr2", 100, 100, 100_150);
index.Add("chr2", 102, 105, 100_200);
index.Flush();
//checking large variants
Assert.Null(index.LargeVariantPositions("chr1", 100, 199));
var largeVariants = index.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]);
}