static public int CreateHmmModelFile(String[] models, String filename)
{
// Generate list of unique accession IDs
List <string> IDs = new List <string>();
foreach (String line in models)
{
String s = line.Trim();
if (s == "")
{
continue;
}
if (s[0] == '>')
{
continue;
}
String id = s.Substring(0, s.IndexOf(':'));
if (!IDs.Contains(id))
{
IDs.Add(id);
}
}
// Parse PFAM-A.HMM and write new HMM file
String fnSource = Hmmer.ModelPath;
StreamReader source = new StreamReader(fnSource);
String fnDest = Hmmer.ModelFolder + filename + ".hmm";
StreamWriter dest = new StreamWriter(fnDest);
int i = 0;
int iWritten = 0;
while (source.Peek() >= 0)
{
String model = Hmmer.ReadNextModel(source);
String acc = Hmmer.GetModelAcc(model);
if (IDs.Contains(acc))
{
dest.Write(model);
iWritten++;
}
i++;
}
if (iWritten != IDs.Count)
{
throw new Exception("Model Count Mismatch");
}
source.Close();
dest.Close();
return(iWritten);
}
static public String FusionHmmScan(String title, String seq1, String seq2)
{
String ret = "";
seq1 = FastaConverter.ConvertString(seq1);
seq2 = FastaConverter.ConvertString(seq2);
List <String> seqs = new List <string>();
for (int i = 0; i < 5; i++)
{
seqs.Add(seq1 + seq2); // + +
seqs.Add(seq1 + Sequencer.InvCompl(seq2)); // + -
seqs.Add(Sequencer.InvCompl(seq1) + seq2); // - +
// 20% deletieren
int cut1 = (int)Math.Round(seq1.Length * 0.2);
int cut2 = (int)Math.Round(seq2.Length * 0.2);
seq1 = seq1.Substring(0, seq1.Length - cut1);
seq2 = seq2.Substring(cut2);
}
List <String> orfs = new List <string>();
foreach (String seq in seqs) // 6-frame ORFs
{
orfs.Add(seq.Substring(0));
orfs.Add(seq.Substring(1));
orfs.Add(seq.Substring(2));
String seqInv = Sequencer.InvCompl(seq);
orfs.Add(seqInv.Substring(0));
orfs.Add(seqInv.Substring(1));
orfs.Add(seqInv.Substring(2));
}
Hmmer hmmer = new Hmmer();
Hmmer.SetDatabase("PFam-A");
Hmmer.Debug = true;
int c = 0;
foreach (String orf in orfs)
{
String peptide = Sequencer.Translate(orf, true); // Translatieren ohne Stop-Codon
hmmer.AddSequence(peptide, title + " #" + c.ToString());
c++;
}
hmmer.DoScan();
List <Hmmer.Result> results = hmmer.GetResults();
List <string> resultsUnique = new List <string>();
ret += "> " + title + Environment.NewLine;
foreach (Hmmer.Result result in results)
{
if (result.Evalue >= 1)
{
continue;
}
string res = result.Accession + ": " + result.Name;
if (!resultsUnique.Contains(res))
{
resultsUnique.Add(res);
ret += res + Environment.NewLine;
}
}
ret += Environment.NewLine;
return(ret);
}
public Result RunSimulations(int numRuns, int numCuts, double probabilityEvent, String modelDatabase = "", bool extendedOutput = false)
{
SimulationId = DateTime.Now.ToString("yyyyMMddhhmmss");
Result res = new Result();
res.SimulationId = SimulationId;
res.ChromosomeName = Source.Name;
res.Runs = numRuns;
res.Cuts = numCuts;
res.Probability = probabilityEvent;
res.Hmm = modelDatabase;
ExtendedOutput = extendedOutput;
App.Log("<div id='chromoDiv" + SimulationId + "'><b>Chromothripsis Simulation ID " + SimulationId + " on Chromosome " + Source.Name + " - " + numRuns.ToString() + " runs " +
"- Simulating " + numCuts.ToString() + " cuts, " + probabilityEvent.ToString() + " probability</b></div>");
App.LogScroll();
DateTime tStart = DateTime.Now;
NumGenes = 0;
NumOrf = 0;
FusionSequences = new List <FusionSequence>();
Hmmer = new Hmmer();
Hmmer.Debug = false;
if (modelDatabase != "")
{
Hmmer.SetDatabase(modelDatabase);
}
if (ExtendedOutput)
{
App.HideLog();
}
for (int i = 0; i < numRuns; i++)
{
Simulate(numCuts, probabilityEvent);
if ((i % 10) == 0)
{
int percent = (int)(((double)i / (double)numRuns) * 100);
App.Status("Simulation Progress: " + percent.ToString() + "%");
}
}
App.Status();
if (ExtendedOutput)
{
App.ShowLog();
}
if (ExtendedOutput)
{
App.Log("");
}
App.Log("Found a total of " + NumGenes.ToString() + " fusion genes<");
if (ExtendedOutput)
{
App.Log("<b>Performing Hmmer Scan...</b>");
App.LogScroll();
}
//App.HideLog();
double totalScore = Hmmer.DoScan();
App.Log("Total Hmmer Score: " + totalScore.ToString("#0.00") + "");
// App.ShowLog();
List <Hmmer.Result> results = Hmmer.GetResults();
int NumHitOrfs = 0;
String resultTable = "";
if (ExtendedOutput)
{
resultTable = "<br/><div id='chromoResults" + SimulationId + "'><table>" +
"<tr><th>Simulation Id</th><th>Sequence ID</th><th>Score</th><th>Target Name</th><th>Accession</th><th>Exon 1</th>" +
"<th>Start 1</th><th>Length 1</th><th>Exon 2</th><th>Start 2</th><th>Length 2</th><th>ORF</th><th>Sequence</th></tr>";
}
foreach (Hmmer.Result result in results)
{
int seqId = Convert.ToInt32(result.SequenceName);
FusionSequence fs = FusionSequences[seqId];
if (!fs.IsHit)
{
fs.IsHit = true;
FusionSequences[seqId] = fs;
NumHitOrfs++;
}
if (ExtendedOutput)
{
resultTable += "<tr><td>" + SimulationId + "</td><td>" + fs.SequenceId + "</td><td>" + result.Score.ToString("#0.00") + "</td><td>" + result.Name + "</td><td>" + result.Accession
+ "</td><td>" + fs.Exon1.Name + "</td><td>" + fs.Range1.Start + "</td><td>" + fs.Range1.Length
+ "</td><td>" + fs.Exon2.Name + "</td><td>" + fs.Range2.Start + "</td><td>" + fs.Range2.Length
+ "</td><td>" + fs.OrfName + "</td><td>" + fs.Sequence +
"</tr>";
}
}
if (ExtendedOutput)
{
resultTable += "</table></div>";
}
App.Log("Number of fusion sequences that produced hits: " + NumHitOrfs + "");
if (ExtendedOutput)
{
App.Log(resultTable);
}
if (ExtendedOutput)
{
App.Log("");
}
DateTime tEnd = DateTime.Now;
TimeSpan tDiff = tEnd - tStart;
double perSecond = ((double)numRuns / (tDiff.TotalMilliseconds / (double)1000));
res.NumHits = results.Count;
res.NumGenes = NumGenes;
res.NumHitOrfs = NumHitOrfs;
res.NumOrfs = NumOrf;
res.TotalScore = totalScore;
res.RunsPerSecond = perSecond;
App.Log("Finished. " + perSecond.ToString("#0") + " runs per second, time elapsed: " + tDiff.ToString(@"mm\:ss"));
App.Log("<br/>");
App.LogScroll();
UpdateOverview(res);
return(res);
}