/// <summary>
/// Load known CN data from a .bed file. File lines have fields:
/// chromosome, start, end, chromcountA, chromcountB
/// So, copy number is the sum of the last 2 fields, major chromosome count is the max of the last 2 fields.
/// </summary>
/// <param name="oracleBedPath"></param>
protected void LoadKnownCNBed(string oracleBedPath)
{
bool stripChr = false;
int count = 0;
this.KnownCN = new Dictionary<string, List<CNInterval>>();
using (StreamReader reader = new StreamReader(oracleBedPath))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null) break;
if (fileLine.Length == 0 || fileLine[0] == '#') continue;
string[] bits = fileLine.Split('\t');
string chromosome = bits[0];
if (stripChr) chromosome = chromosome.Replace("chr", "");
if (!KnownCN.ContainsKey(chromosome)) KnownCN[chromosome] = new List<CNInterval>();
CNInterval interval = new CNInterval();
interval.Start = int.Parse(bits[1]);
interval.End = int.Parse(bits[2]);
interval.CN = int.Parse(bits[3]) + int.Parse(bits[4]);
KnownCN[chromosome].Add(interval);
count++;
}
}
Console.WriteLine(">>>Loaded {0} known-CN intervals", count);
}
/// <summary>
/// Load known CN data from a .bed file. File lines have fields:
/// chromosome, start, end, chromcountA, chromcountB
/// So, copy number is the sum of the last 2 fields, major chromosome count is the max of the last 2 fields.
/// </summary>
/// <param name="oracleBedPath"></param>
protected void LoadKnownCNBed(string oracleBedPath)
{
bool stripChr = false;
int count = 0;
this.KnownCN = new Dictionary <string, List <CNInterval> >();
using (FileStream stream = new FileStream(oracleBedPath, FileMode.Open, FileAccess.Read))
using (StreamReader reader = new StreamReader(stream))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null)
{
break;
}
if (fileLine.Length == 0 || fileLine[0] == '#')
{
continue;
}
string[] bits = fileLine.Split('\t');
string chromosome = bits[0];
if (stripChr)
{
chromosome = chromosome.Replace("chr", "");
}
if (!KnownCN.ContainsKey(chromosome))
{
KnownCN[chromosome] = new List <CNInterval>();
}
CNInterval interval = new CNInterval();
interval.Start = int.Parse(bits[1]);
interval.End = int.Parse(bits[2]);
interval.CN = int.Parse(bits[3]) + int.Parse(bits[4]);
if (bits.Length > 5)
{
interval.Heterogeneity = double.Parse(bits[5]);
}
else
{
interval.Heterogeneity = -1.0;
}
KnownCN[chromosome].Add(interval);
count++;
}
}
Console.WriteLine(">>>Loaded {0} known-CN intervals", count);
}
protected void LoadKnownCNVCF(string oracleVCFPath)
{
bool stripChr = false;
// Load our "oracle" of known copy numbers:
this.KnownCN = new Dictionary <string, List <CNInterval> >();
int count = 0;
using (GzipReader reader = new GzipReader(oracleVCFPath))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null)
{
break;
}
if (fileLine.Length == 0 || fileLine[0] == '#')
{
continue;
}
string[] bits = fileLine.Split('\t');
if (bits.Length == 1 && bits[0].Trim().Length == 0)
{
continue; // skip empty lines!
}
string chromosome = bits[0];
if (stripChr)
{
chromosome = chromosome.Replace("chr", "");
}
if (!KnownCN.ContainsKey(chromosome))
{
KnownCN[chromosome] = new List <CNInterval>();
}
CNInterval interval = new CNInterval();
interval.Start = int.Parse(bits[1]);
interval.CN = -1;
string[] infoBits = bits[7].Split(';');
foreach (string subBit in infoBits)
{
if (subBit.StartsWith("CN="))
{
float tempCN = float.Parse(subBit.Substring(3));
if (subBit.EndsWith(".5"))
{
interval.CN = (int)Math.Round(tempCN + 0.1); // round X.5 up to X+1
}
else
{
interval.CN = (int)Math.Round(tempCN); // Round off
}
}
if (subBit.StartsWith("END="))
{
interval.End = int.Parse(subBit.Substring(4));
}
}
// Parse CN from Canvas output:
if (bits.Length > 8)
{
string[] subBits = bits[8].Split(':');
string[] subBits2 = bits[9].Split(':');
for (int subBitIndex = 0; subBitIndex < subBits.Length; subBitIndex++)
{
if (subBits[subBitIndex] == "CN")
{
interval.CN = int.Parse(subBits2[subBitIndex]);
}
}
}
if (interval.End == 0 || interval.CN < 0)
{
Console.WriteLine("Error - bogus record!");
Console.WriteLine(fileLine);
}
else
{
KnownCN[chromosome].Add(interval);
count++;
}
}
}
Console.WriteLine(">>>Loaded {0} known-CN intervals", count);
}
protected void LoadKnownCNVCF(string oracleVCFPath)
{
bool stripChr = false;
// Load our "oracle" of known copy numbers:
this.KnownCN = new Dictionary<string, List<CNInterval>>();
int count = 0;
using (GzipReader reader = new GzipReader(oracleVCFPath))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null) break;
if (fileLine.Length == 0 || fileLine[0] == '#') continue;
string[] bits = fileLine.Split('\t');
if (bits.Length == 1 && bits[0].Trim().Length == 0) continue; // skip empty lines!
string chromosome = bits[0];
if (stripChr) chromosome = chromosome.Replace("chr", "");
if (!KnownCN.ContainsKey(chromosome)) KnownCN[chromosome] = new List<CNInterval>();
CNInterval interval = new CNInterval();
interval.Start = int.Parse(bits[1]);
interval.CN = -1;
string[] infoBits = bits[7].Split(';');
foreach (string subBit in infoBits)
{
if (subBit.StartsWith("CN="))
{
float tempCN = float.Parse(subBit.Substring(3));
if (subBit.EndsWith(".5"))
{
interval.CN = (int)Math.Round(tempCN + 0.1); // round X.5 up to X+1
}
else
{
interval.CN = (int)Math.Round(tempCN); // Round off
}
}
if (subBit.StartsWith("END="))
{
interval.End = int.Parse(subBit.Substring(4));
}
}
// Parse CN from Canvas output:
if (bits.Length > 8)
{
string[] subBits = bits[8].Split(':');
string[] subBits2 = bits[9].Split(':');
for (int subBitIndex = 0; subBitIndex < subBits.Length; subBitIndex++)
{
if (subBits[subBitIndex] == "CN")
{
interval.CN = int.Parse(subBits2[subBitIndex]);
}
}
}
if (interval.End == 0 || interval.CN < 0)
{
Console.WriteLine("Error - bogus record!");
Console.WriteLine(fileLine);
}
else
{
KnownCN[chromosome].Add(interval);
count++;
}
}
}
Console.WriteLine(">>>Loaded {0} known-CN intervals", count);
}
/// <summary>
/// Developer debug method:
/// - Split each truth interval to have at least the same segmentation as the called segments
/// (Note: We are intentionally ignoring segments - or parts thereof - called in areas not defined in the Truth set)
/// - For each QScore method:
/// - Report these new intervals and associated QScores to an extended report output file
/// - Generate ROC output
/// </summary>
private void GenerateExtendedReportVersusKnownCN()
{
Dictionary<string, List<CNInterval>> resegmentedKnownCN = new Dictionary<string, List<CNInterval>>();
// Copy KnownCN entries to working container
foreach (string chr in this.CNOracle.KnownCN.Keys)
{
resegmentedKnownCN[chr] = new List<CNInterval>();
foreach (CNInterval interval in this.CNOracle.KnownCN[chr])
{
CNInterval newInterval = new CNInterval();
newInterval.Start = interval.Start;
newInterval.End = interval.End;
newInterval.CN = interval.CN;
resegmentedKnownCN[chr].Add(newInterval);
}
}
// Split each truth interval to match the segments' breakpoints
foreach (CanvasSegment segment in this.Segments)
{
if (!resegmentedKnownCN.ContainsKey(segment.Chr)) continue;
for (int i = 0; i < resegmentedKnownCN[segment.Chr].Count; i++) // Using for loop instead of foreach because we add items to the list
{
CNInterval interval = resegmentedKnownCN[segment.Chr][i];
if (interval.Start == segment.Begin && interval.End == segment.End) break; // perfect segment-knownCN match
if (interval.Start >= segment.End || interval.End <= segment.Begin) continue; // segment completely outside this knownCN
// If necessary, split interval at segment.Begin position (i.e. extract sub-interval preceding segment)
if (segment.Begin > interval.Start)
{
CNInterval newInterval = new CNInterval();
newInterval.Start = interval.Start;
newInterval.End = segment.Begin;
newInterval.CN = interval.CN;
interval.Start = newInterval.End;
resegmentedKnownCN[segment.Chr].Add(newInterval);
}
// If necessary, split interval at segment.End position (i.e. extract sub-interval following segment)
if (segment.End < interval.End)
{
CNInterval newInterval = new CNInterval();
newInterval.Start = segment.End;
newInterval.End = interval.End;
newInterval.CN = interval.CN;
interval.End = newInterval.Start;
resegmentedKnownCN[segment.Chr].Add(newInterval);
}
}
}
// Sort list of new intervals by starting position, just for prettiness
foreach (List<CNInterval> list in resegmentedKnownCN.Values)
{
list.Sort((i1, i2) => i1.Start.CompareTo(i2.Start));
}
// Generate ROC output data for each QScore method
foreach (CanvasSegment.QScoreMethod qscoreMethod in CanvasSegment.QScoreMethod.GetValues(typeof(CanvasSegment.QScoreMethod)))
{
GenerateReportAndRocDataForQscoreMethod(qscoreMethod, resegmentedKnownCN);
}
}
