public override void Run(string analysisName, UpdateProgressBar updateProgressBar)
{
_dataManager.PercentIdentities[analysisName] = new PercentIdentityData();
foreach (KeyValuePair <string, string> sequence1 in _dataManager.Alignments[analysisName].Contents)
{
_dataManager.PercentIdentities[analysisName].Dic[sequence1.Key] = new Dictionary <string, float>();
_dataManager.PercentIdentities[analysisName].DicInverse[sequence1.Key] = new Dictionary <string, float>();
if (!_dataManager.SequencesUsed[sequence1.Key].Vaccine)
{
_dataManager.PercentIdentities[analysisName].Sequences.Add(sequence1.Key);
}
List <float> dataList = new List <float>();
List <float> vdata = new List <float>();
float total = 0;
int count = 0;
Dictionary <int, Dictionary <string, string> > siteChanges = new Dictionary <int, Dictionary <string, string> >();
foreach (KeyValuePair <string, string> sequence2 in _dataManager.Alignments[analysisName].Contents)
{
float percent = GlobalCalculations.CalculatePercentIdentity(sequence1.Value, sequence2.Value, sequence2.Key, siteChanges);
if (sequence1.Key != sequence2.Key)
{
total += percent;
}
count++;
if (!_dataManager.PercentIdentities[analysisName].Dic[sequence1.Key].ContainsKey(sequence2.Key))
{
_dataManager.PercentIdentities[analysisName].Dic[sequence1.Key][sequence2.Key] = percent;
_dataManager.PercentIdentities[analysisName].DicInverse[sequence1.Key][sequence2.Key] = 100 - percent;
}
dataList.Add(percent);
if (!_dataManager.SequencesUsed[sequence2.Key].Vaccine)
{
vdata.Add(percent);
}
}
_dataManager.PercentIdentities[analysisName].SiteChanges[sequence1.Key] = siteChanges;
if (!_dataManager.SequencesUsed[sequence1.Key].Vaccine)
{
_dataManager.PercentIdentities[analysisName].Data.Add(vdata);
}
}
updateProgressBar((int)(20 / (float)_dataManager.AnalysisCount));
}
/// <summary>
/// Compares all orfs to the orf templates to find all of the known orfs
/// </summary>
/// <param name="allOrfs"></param>
/// <returns>Dictionary of orfs with location data</returns>
/*private Dictionary<string, OrfData> findKnownOrfs(Dictionary<string, OrfData> allOrfs)
* {
* Dictionary<string, OrfData> knownOrfs = new Dictionary<string, OrfData>();
* Dictionary<string, List<OrfData>> potentialOrfs = new Dictionary<string, List<OrfData>>();
* OrfsTemplate orfTemplates = _dataManager.OrfTemplates[_dataManager.CurrentVirusKey];
* foreach(OrfTemplate orfTemplate in orfTemplates.Orfs)
* {
* // Find potential orfs
* foreach(KeyValuePair<string, OrfData> allOrfsPair in allOrfs)
* {
* int lengthBuffer = Convert.ToInt32(orfTemplate.LengthAA * _dataManager.OrfLengthThreshold);
* if(allOrfsPair.Value.LengthAA > orfTemplate.LengthAA - lengthBuffer &&
* allOrfsPair.Value.LengthAA < orfTemplate.LengthAA + lengthBuffer)
* {
* if (_dataManager.PartialOrfFile || (allOrfsPair.Value.StartLocationN - _dataManager.OrfSiteRange < orfTemplate.StartSite &&
* allOrfsPair.Value.StartLocationN + _dataManager.OrfSiteRange > orfTemplate.StartSite)){
* if (!potentialOrfs.ContainsKey(orfTemplate.Name)) potentialOrfs.Add(orfTemplate.Name, new List<OrfData>());
* potentialOrfs[orfTemplate.Name].Add(allOrfsPair.Value);
* }
* }
* }
*
* // Get hardset
* if (orfTemplate.HardSet && potentialOrfs.ContainsKey(orfTemplate.Name))
* {
* for(int i = 0; i < potentialOrfs[orfTemplate.Name].Count; i++)
* {
* potentialOrfs[orfTemplate.Name][i].StartLocationN = potentialOrfs[orfTemplate.Name][i].EndLocationN - orfTemplate.LengthAA * 3;
* potentialOrfs[orfTemplate.Name][i].StartLocationAA = potentialOrfs[orfTemplate.Name][i].EndLocationAA - orfTemplate.LengthAA;
* potentialOrfs[orfTemplate.Name][i].LengthAA = orfTemplate.LengthAA;
* potentialOrfs[orfTemplate.Name][i].LengthN = orfTemplate.LengthAA * 3;
* }
* }
*
* // Find closest related orf
* bool found = false;
* KeyValuePair<float, OrfData> highestOrf = new KeyValuePair<float, OrfData>(_dataManager.OrfIdentifierPIThreshold, new OrfData());
* if (potentialOrfs.ContainsKey(orfTemplate.Name))
* {
* float pi = 0;
* foreach (OrfData potentialOrfPair in potentialOrfs[orfTemplate.Name])
* {
* pi = GlobalCalculations.CalculatePercentIdentity(potentialOrfPair.ContentsAA, orfTemplate.Sequence);
* if (pi > highestOrf.Key)
* {
* highestOrf = new KeyValuePair<float, OrfData>(pi, potentialOrfPair);
* found = true;
* }
* }
* if (found) {
* knownOrfs.Add(orfTemplate.Name, highestOrf.Value);
* knownOrfs[orfTemplate.Name].Name = orfTemplate.Name;
* if (!_dataManager.AnalysisNames.Contains(orfTemplate.Name + "_n"))
* _dataManager.AnalysisNames.Add(orfTemplate.Name + "_n");
* if (!_dataManager.AnalysisNames.Contains(orfTemplate.Name + "_aa"))
* _dataManager.AnalysisNames.Add(orfTemplate.Name + "_aa");
* }
* }
* }
* return knownOrfs;
* }*/
private Dictionary <string, int[]> findOrfAlignmentLocations(string relatedSequence)
{
Dictionary <string, int[]> locations = new Dictionary <string, int[]>();
OrfsTemplate orfTemplates = _dataManager.OrfTemplates[_dataManager.CurrentVirusKey];
string contents = _dataManager.Alignments["Wholegenome"].Contents[relatedSequence];
Dictionary <string, int[]> allOrfs = findAllOrfs(contents);
foreach (OrfTemplate orfTemplate in orfTemplates.Orfs)
{
int[] closestOrf = new int[3];
float highestPI = 0;
foreach (int[] orf in allOrfs.Values)
{
int start = orf[0];
if (orfTemplate.HardSet)
{
start = findStartPos(contents.Replace("-", "").Substring(orfTemplate.StartSite - 1, orfTemplate.LengthAA * 3), contents) - 2;
if (start > orf[1])
{
start = 0;
}
}
string newOrfContents = NucleotideToAminoAcid(contents.Substring(start, orf[1] - start).Replace("-", ""));
float pi = GlobalCalculations.CalculatePercentIdentity(newOrfContents, orfTemplate.Sequence);
if (pi > highestPI)
{
highestPI = pi; closestOrf = new int[3] {
start, orf[1], orf[2]
};
}
}
locations[orfTemplate.Name] = closestOrf;
if (!_dataManager.AnalysisNames.Contains(orfTemplate.Name + "_n"))
{
_dataManager.AnalysisNames.Add(orfTemplate.Name + "_n");
}
if (!_dataManager.AnalysisNames.Contains(orfTemplate.Name + "_aa"))
{
_dataManager.AnalysisNames.Add(orfTemplate.Name + "_aa");
}
}
return(locations);
}