/// <summary>
///
/// </summary>
public void UpdateChainInterfaceQscores()
{
ProtCidSettings.dataType = "pfam";
DbBuilderHelper.Initialize();
ProtCidSettings.progressInfo.Reset();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating Q scores of SuperGroups");
ChainGroupRepEntryComp interGroupRepEntryComp = new ChainGroupRepEntryComp();
interGroupRepEntryComp.CompareMissingEntryPairs();
/*
* ProtCidSettings.logWriter.WriteLine(DateTime.Today.ToShortTimeString());
* ProtCidSettings.logWriter.WriteLine("Updating Q scores of chain interfaces in chain groups.");
* ProtCidSettings.logWriter.Flush();
*
* Hashtable updatedGroupHash = GetUpdatedGroups();
* Hashtable updateSuperGroupHash = null;
*
* if (updateSuperGroupHash == null)
* {
* updateSuperGroupHash = new Hashtable();
* StreamReader dataReader = new StreamReader("UpdateSuperGroups.txt");
* string line = "";
* int superGroupId = -1;
* Hashtable updateGroupEntryHash = null;
* while ((line = dataReader.ReadLine()) != null)
* {
* if (line == "")
* {
* continue;
* }
* if (line.Substring(0, 1) == "#")
* {
* if (superGroupId != -1)
* {
* updateSuperGroupHash.Add(superGroupId, updateGroupEntryHash);
* }
* superGroupId = Convert.ToInt32(line.Substring(1, line.Length - 1));
* updateGroupEntryHash = new Hashtable();
* }
* if (line.IndexOf(":") > -1)
* {
* string[] fields = line.Split(':');
* string[] entries = fields[1].Split(',');
* updateGroupEntryHash.Add(Convert.ToInt32(fields[0]), entries);
* }
* }
* if (superGroupId > -1)
* {
* updateSuperGroupHash.Add(superGroupId, updateGroupEntryHash);
* }
* dataReader.Close();
* }
* ArrayList updateSuperGroupList = new ArrayList(updateSuperGroupHash.Keys);
* updateSuperGroupList.Sort();
* int[] updateSuperGroups = new int[updateSuperGroupList.Count];
* updateSuperGroupList.CopyTo(updateSuperGroups);
*
* // number of calculations in the order: 189798, 2812822, 331348, 287770, 102644, 372940, 124384
* int[] excludedChainGroups = { 7, 744, 762, 1379, 1620, 4116, 4949 };
*
* ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating the interface comparisons between groups.");
* ChainGroupRepEntryComp interGroupRepEntryComp = new ChainGroupRepEntryComp();
* // interGroupRepEntryComp.UpdateEntryComparisonInSuperGroups(updateSuperGroupHash);
* int bigChainGroupId = 4949;
* interGroupRepEntryComp.UpdateEntryComparisonInSuperGroup (updateSuperGroupHash, bigChainGroupId);*/
}
/// <summary>
///
/// </summary>
public void UpdateChainInterfaceClusters()
{
ProtCidSettings.dataType = "pfam";
DbBuilderHelper.Initialize();
ProtCidSettings.progressInfo.Reset();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating PFAM SuperGroup Interfaces Data.");
// ChainClusterStat clusterStat = new ChainClusterStat();
// clusterStat.UpdateMinSeqIdentityInDb();
ProtCidSettings.logWriter.WriteLine(DateTime.Today.ToShortTimeString());
ProtCidSettings.logWriter.WriteLine("Updating PFAM SuperGroup Interfaces Data.");
ProtCidSettings.logWriter.Flush();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Update biological units and entrypfamarch data for protcid");
Dictionary <int, string[]> updatedGroupHash = GetUpdatedGroups();
Dictionary <int, Dictionary <int, string[]> > updateSuperGroupHash = null;
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating Super Groups.");
// ChainGroupsClassifier superGroupClassifier = new ChainGroupsClassifier();
// updateSuperGroupHash = superGroupClassifier.UpdateFamilySuperGroups(updatedGroupHash);
if (updateSuperGroupHash == null)
{
updateSuperGroupHash = new Dictionary <int, Dictionary <int, string[]> > ();
StreamReader dataReader = new StreamReader("UpdateSuperGroups.txt");
string line = "";
int superGroupId = -1;
Dictionary <int, string[]> updateGroupEntryHash = null;
while ((line = dataReader.ReadLine()) != null)
{
if (line == "")
{
continue;
}
if (line.Substring(0, 1) == "#")
{
if (superGroupId != -1)
{
updateSuperGroupHash.Add(superGroupId, updateGroupEntryHash);
}
superGroupId = Convert.ToInt32(line.Substring(1, line.Length - 1));
updateGroupEntryHash = new Dictionary <int, string[]> ();
}
if (line.IndexOf(":") > -1)
{
string[] fields = line.Split(':');
string[] entries = fields[1].Split(',');
updateGroupEntryHash.Add(Convert.ToInt32(fields[0]), entries);
}
}
if (superGroupId > -1)
{
updateSuperGroupHash.Add(superGroupId, updateGroupEntryHash);
}
dataReader.Close();
}
List <int> updateSuperGroupList = new List <int> (updateSuperGroupHash.Keys);
updateSuperGroupList.Sort();
int[] updateSuperGroups = new int[updateSuperGroupList.Count];
updateSuperGroupList.CopyTo(updateSuperGroups);
string[] updateEntries = GetUpdateEntries(updateSuperGroupHash);
// clear the any existing file to make sure the file containing only those from
// the comparison between groups.
string nonAlignedPairFile = "NonAlignedEntryPairs.txt";
if (File.Exists(nonAlignedPairFile))
{
// before delete, make a copy
File.Copy(nonAlignedPairFile, Path.Combine(ProtCidSettings.dirSettings.fatcatPath, nonAlignedPairFile), true);
File.Delete(nonAlignedPairFile);
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating the interface comparisons between groups.");
ChainGroupRepEntryComp interGroupRepEntryComp = new ChainGroupRepEntryComp();
interGroupRepEntryComp.UpdateEntryComparisonInSuperGroups(updateSuperGroupHash);
// interGroupRepEntryComp.CompareMissingEntryPairs();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating super groups clustering.");
ChainInterfaceCluster interfaceCluster = new ChainInterfaceCluster();
interfaceCluster.UpdateSuperGroupClusters(updateSuperGroups);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating the summary data.");
ChainClusterStat superClusterStat = new ChainClusterStat();
superClusterStat.UpdateSupergroupInterfaceClustersSumInfo(updateSuperGroups, updateEntries, "pfam", true);
// superClusterStat.UpdateProtCidChainMetaData("pfam", updateEntries, updateSuperGroups);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating tar interface files.");
InterfaceFilesReverser interfaceFileReverse = new InterfaceFilesReverser();
string updateClusterReverseFileList = interfaceFileReverse.ReverseClusterInterfaceFiles(updateSuperGroupHash);
// string updateClusterReverseFileList = "ReverseInterfacesInCluster.txt";
updateClusterReverseFileList = Path.Combine(ProtCidSettings.applicationStartPath, updateClusterReverseFileList);
// string updateClusterReverseFileList = "ReverseInterfacesInCluster.txt";
ChainClusterCompress clusterFileCompress = new ChainClusterCompress();
clusterFileCompress.CompressGroupClusterInterfaceFiles(updateSuperGroups, updateClusterReverseFileList);
clusterFileCompress.UpdateCrystInterfaceFilesNotInClusters(updateSuperGroups, false);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating sequence files.");
ProtCidSettings.logWriter.WriteLine("Updating sequence files.");
SeqFastaGenerator seqFastaGen = new SeqFastaGenerator();
seqFastaGen.UpdateSeqFastaFiles(updateSuperGroups);
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.logWriter.Flush();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Update biological units and entrypfamarch data for protcid");
ProtCidSettings.logWriter.WriteLine("Update biological units and entrypfamarch data for protcid");
BiolUnitSumInfo buSumInfo = new BiolUnitSumInfo();
buSumInfo.UpdateBiolUnits(updateEntries);
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.logWriter.Flush();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating cluster interface images.");
ProtCidSettings.logWriter.WriteLine("Updating cluster interface images.");
InterfaceImageGen imageGen = new InterfaceImageGen();
imageGen.UpdateClusterInterfaceImages(updateSuperGroups);
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.logWriter.Flush();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating indexes for super groups tables");
ProtCidSettings.logWriter.WriteLine("Updating indexes for super groups tables");
DbBuilderHelper.UpdateIndexes("PfamSuper", ProtCidSettings.protcidDbConnection);
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.logWriter.Flush();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Update Done!");
ProtCidSettings.progressInfo.threadFinished = true;
ProtCidSettings.logWriter.Flush();
}
public void BuildChainInterfaceClusters(int operationStep)
{
ProtCidSettings.dataType = "pfam";
DbBuilderHelper.Initialize();
switch (operationStep)
{
case 1:
try
{
ChainGroupsClassifier chainGroupClassifier = new ChainGroupsClassifier();
chainGroupClassifier.CombineFamilyGroups();
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Classify chain relations error: " +
ex.Message);
ProtCidSettings.logWriter.WriteLine("Classify chain relations error: " +
ex.Message);
ProtCidSettings.logWriter.Flush();
}
break;
// goto case 2;
case 2:
ChainGroupRepEntryComp interGroupRepEntryComp = new ChainGroupRepEntryComp();
interGroupRepEntryComp.CompareTwoEntries("5uvz", "2dww");
interGroupRepEntryComp.CompareInterGroupRepEntryInterfacesInChainGroups();
// interGroupRepEntryComp.PrintMissingAlignInterGroupRepEntries();
Alignments.GroupEntryAlignments groupAlign = new Alignments.GroupEntryAlignments();
groupAlign.GetRepChainPairsToBeAligned();
// interGroupRepEntryComp.CompareMissingEntryPairs();
break;
// goto case 3;
case 3:
ChainInterfaceCluster interfaceCluster = new ChainInterfaceCluster();
interfaceCluster.ClusterChainGroupInterfaces();
/* int[] abSuperGroupIds = { 744};
* interfaceCluster.UpdateSuperGroups(abSuperGroupIds); */
break;
// goto case 4;
case 4:
ChainClusterStat superClusterStat = new ChainClusterStat();
// superClusterStat.PrintChainClusterTextFilesFromDb ();
superClusterStat.PrintSupergroupInterfaceClusters("pfam");
// goto case 5;
break;
case 5:
InterfaceFilesReverser interfaceFileReverse = new InterfaceFilesReverser();
// interfaceFileReverse.UpdateIsSymmetry();
string clusterReverseFileList = interfaceFileReverse.ReverseClusterInterfaceFiles();
#if DEBUG
/////////////////////////////////////
// the clusterReverseFileList is still a problem, file not exist exception, probably due to change the current directory
// but I can not detect where the problem is.
// should fix it.
////////////////////////////////////
ProtCidSettings.progressInfo.progStrQueue.Enqueue("The current working directory: " + System.Environment.CurrentDirectory);
ProtCidSettings.logWriter.WriteLine("The current working directory: " + System.Environment.CurrentDirectory);
FileInfo fileInfo = new FileInfo(clusterReverseFileList);
ProtCidSettings.logWriter.WriteLine("The directory of clusterReverseFileList file: " + fileInfo.DirectoryName);
#endif
clusterReverseFileList = Path.Combine(ProtCidSettings.applicationStartPath, clusterReverseFileList);
ChainClusterCompress clusterFileCompress = new ChainClusterCompress();
clusterFileCompress.CompressClusterInterfaceFiles(clusterReverseFileList);
// copy the interface files for those not in any clusters
// save disk space in the web server
clusterFileCompress.RetrieveCrystInterfaceFilesNotInClusters(false);
InterfaceImageGen imageGen = new InterfaceImageGen();
imageGen.GenerateInterfaceImages();
// goto case 6;
break;
case 6:
SeqFastaGenerator seqFastaGen = new SeqFastaGenerator();
seqFastaGen.WriteSequencesToFastaFiles();
break;
// goto case 7;
case 7:
BiolUnitSumInfo buSumInfo = new BiolUnitSumInfo();
// buSumInfo.UpdateBiolUnits();
buSumInfo.RetrieveBiolUnits();
break;
default:
break;
}
DbBuilderHelper.UpdateIndexes("PfamSuper", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.threadFinished = true;
}
/// <summary>
///
/// </summary>
/// <param name="userGroupName"></param>
/// <param name="userGroupSeqId"></param>
/// <param name="userEntries"></param>
public void FindInterfaceClustersInUserGroup(string userGroupName, int userGroupSeqId, string[] userEntries)
{
ProtCidSettings.progressInfo.Reset();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Clutering interfaces in the user group");
ProtCidSettings.progressInfo.progStrQueue.Enqueue(userGroupName + " " + userGroupSeqId.ToString());
Dictionary <int, string[]> updatedGroupHash = GetUpdatedGroups(userEntries);
Dictionary <int, Dictionary <int, string[]> > updateSuperGroupHash = new Dictionary <int, Dictionary <int, string[]> > ();
updateSuperGroupHash.Add(userGroupSeqId, updatedGroupHash);
Dictionary <int, string> superGroupNameHash = new Dictionary <int, string> ();
superGroupNameHash.Add(userGroupSeqId, userGroupName);
List <int> updateSuperGroupList = new List <int> (updateSuperGroupHash.Keys);
int[] updateSuperGroups = new int[updateSuperGroupList.Count];
updateSuperGroupList.CopyTo(updateSuperGroups);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating Super Groups.");
ChainGroupsClassifier superGroupClassifier = new ChainGroupsClassifier();
superGroupClassifier.AddChainRelGroupsBasedOnPfam(updateSuperGroupHash, superGroupNameHash);
// clear the any existing file to make sure the file containing only those from
// the comparison between groups.
string nonAlignedPairFile = "NonAlignedEntryPairs.txt";
if (File.Exists(nonAlignedPairFile))
{
// before delete, make a copy
File.Copy(nonAlignedPairFile, Path.Combine(ProtCidSettings.dirSettings.fatcatPath, nonAlignedPairFile));
File.Delete(nonAlignedPairFile);
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating the interface comparisons between groups.");
ChainGroupRepEntryComp interGroupRepEntryComp = new ChainGroupRepEntryComp();
interGroupRepEntryComp.UpdateEntryComparisonInSuperGroups(updateSuperGroupHash);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating super groups clustering.");
ChainInterfaceCluster interfaceCluster = new ChainInterfaceCluster();
interfaceCluster.ClusterUserDefinedGroupInterfaces(updateSuperGroups);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating the summary data.");
ChainClusterStat superClusterStat = new ChainClusterStat();
superClusterStat.UpdateSupergroupInterfaceClustersSumInfo(updateSuperGroups, userEntries, "pfam", false);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating tar interface files.");
InterfaceFilesReverser interfaceFileReverse = new InterfaceFilesReverser();
string updateClusterReverseListFile = interfaceFileReverse.ReverseClusterInterfaceFiles(updateSuperGroupHash);
// string updateClusterReverseListFile = "ReverseInterfacesInCluster.txt";
ChainClusterCompress clusterFileCompress = new ChainClusterCompress();
clusterFileCompress.CompressGroupClusterInterfaceFiles(updateSuperGroups, updateClusterReverseListFile);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating sequence files.");
SeqFastaGenerator seqFastaGen = new SeqFastaGenerator();
seqFastaGen.UpdateSeqFastaFiles(updateSuperGroups);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
}