/// <summary>
/// update entry-related data
/// it is not necessary to recompute entry data from scratch
/// </summary>
/// <param name="updateEntries"></param>
public void UpdateEntryInterfaceData(string[] updateEntries)
{
DbBuilderHelper.Initialize();
if (updateEntries == null)
{
// string lsFile = @"D:\Qifang\ProjectData\DbProjectData\PDB\newls-pdb_entry.txt";
// updateEntries = GetUpdateEntries(lsFile);
updateEntries = GetUpdateEntries();
}
ProtCidSettings.logWriter.WriteLine("1. Compute interfaces from crystals");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("1. Compute interfaces from crystals");
CrystInterfaceBuilder interfaceBuilder = new CrystInterfaceBuilder();
interfaceBuilder.ComputeEntryCrystInterfaces(updateEntries);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
ProtCidSettings.logWriter.WriteLine("2. Generate interface files, calculate ASA and contacts in interfaces");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("2. Generate interface files, calculate ASA and contacts in interfaces");
interfaceBuilder.GenerateEntryInterfaceFiles(updateEntries);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
ProtCidSettings.logWriter.WriteLine("3. Compare cryst interfaces with PDB/PISA BU and ASU Interfaces");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("3. Compare cryst interfaces with PDB/PISA BU and ASU Interfaces");
// compare cryst interfaces and BU interfaces for each entry
EntryCrystBuInterfaceComp crystBuInterfaceComp = new EntryCrystBuInterfaceComp();
crystBuInterfaceComp.UpdateEntryCrystBuInterfaces(updateEntries);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.logWriter.WriteLine("4. Update crc-crc hh alignments");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("4. Update crc-crc hh alignments");
Alignments.GroupEntryAlignments groupAlignment = new InterfaceClusterLib.Alignments.GroupEntryAlignments();
groupAlignment.UpdateCrcHhAlignments();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("CrystBuInterfaceComp", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.logWriter.WriteLine("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
}
public void PrintNonAlignedPairsFile()
{
DbBuilderHelper.Initialize();
StreamReader dataReader = new StreamReader("RepEntriesInterGroupCompLog.txt");
StreamWriter dataWriter = new StreamWriter("NonAlignedPairs.txt");
StreamWriter noInterfaceEntryWriter = new StreamWriter("NoInterfaceEntries.txt");
string line = "";
List <string> noInterfaceEntryList = new List <string> ();
while ((line = dataReader.ReadLine()) != null)
{
if (line.IndexOf("not aligned pair") > -1)
{
string[] fields = line.Split(' ');
string[] entityChains1 = GetPolypeptideChains(fields[1]);
string[] entityChains2 = GetPolypeptideChains(fields[3]);
foreach (string entityChain1 in entityChains1)
{
foreach (string entityChain2 in entityChains2)
{
dataWriter.WriteLine(fields[1] + entityChain1 + "\t" + fields[3] + entityChain2);
}
}
}
else
{
int index = line.IndexOf("no cryst interfaces");
if (index > -1)
{
string pdbId = line.Substring(index - 5, 4);
if (!noInterfaceEntryList.Contains(pdbId))
{
noInterfaceEntryList.Add(pdbId);
noInterfaceEntryWriter.WriteLine(pdbId);
}
}
}
}
dataReader.Close();
dataWriter.Close();
noInterfaceEntryWriter.Close();
}
private SchemaObjectCollection LoadAURolesInner(string[] unitIds, string[] schemaRoleIDs, bool normalOnly, DateTime timePoint)
{
string sql = @"SELECT O.* FROM SC.SchemaObject O INNER JOIN SC.AURoleSnapshot S
ON O.ID = S.ID AND O.VersionStartTime = S.VersionStartTime AND O.SchemaType = S.SchemaType
INNER JOIN SC.SchemaMembers R ON O.ID = R.MemberID AND O.SchemaType = R.MemberSchemaType
WHERE";
IConnectiveSqlClause inBuilder = DbBuilderHelper.In("S.SchemaRoleID", schemaRoleIDs);
IConnectiveSqlClause inBuilder2 = DbBuilderHelper.In("R.ContainerID", unitIds);
WhereSqlClauseBuilder where = new WhereSqlClauseBuilder();
if (normalOnly) where.NormalFor("O.Status").NormalFor("R.Status");
var timeCondition1 = VersionStrategyQuerySqlBuilder.Instance.TimePointToBuilder(timePoint, "S.");
var timeCondition2 = VersionStrategyQuerySqlBuilder.Instance.TimePointToBuilder(timePoint, "O.");
sql += new ConnectiveSqlClauseCollection(timeCondition1, timeCondition2, where, inBuilder, inBuilder2).ToSqlString(TSqlBuilder.Instance);
return LoadSchemaObjects(sql);
}
/// <summary>
///
/// </summary>
public void UpdatePfamChainClusterFiles()
{
ProtCidSettings.dataType = "pfam";
DbBuilderHelper.Initialize();
ProtCidSettings.progressInfo.Reset();
// int[] updateSuperGroups = GetUpdateClusterChainGroups ();
int[] updateSuperGroups = { 3171, 9125, 13844, 16380, 28272, 29578 };
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, "pfam", true);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating tar interface files.");
InterfaceFilesReverser interfaceFileReverse = new InterfaceFilesReverser();
string updateClusterReverseFileList = interfaceFileReverse.ReverseClusterInterfaceFiles(updateSuperGroups);
string updateClusterReverseFileList1 = "ReverseInterfacesInCluster.txt";
ChainClusterCompress clusterFileCompress = new ChainClusterCompress();
clusterFileCompress.CompressGroupClusterInterfaceFiles(updateSuperGroups, updateClusterReverseFileList1);
clusterFileCompress.UpdateCrystInterfaceFilesNotInClusters(false);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating sequence files.");
SeqFastaGenerator seqFastaGen = new SeqFastaGenerator();
seqFastaGen.UpdateSeqFastaFiles(updateSuperGroups);
InterfaceImageGen imageGen = new InterfaceImageGen();
imageGen.UpdateClusterInterfaceImages(updateSuperGroups);
}
public void ClusterPkinaseInterfaces()
{
ProtCidSettings.dataType = "pfam";
DbBuilderHelper.Initialize();
string pfamId = "";
string[] pkinaseEntries = null;
string groupArchString = "";
int userGroupSeqId = 111111;
pfamId = "Pkinase";
pkinaseEntries = GetPfamEntries(pfamId);
groupArchString = pfamId;
userGroupSeqId = 111111;
chainRelInterfaceBuilder.FindInterfaceClustersInUserGroup(groupArchString, userGroupSeqId, pkinaseEntries);
pfamId = "Pkinase_Tyr";
pkinaseEntries = GetPfamEntries(pfamId);
groupArchString = pfamId;
userGroupSeqId = userGroupSeqId + 1;
chainRelInterfaceBuilder.FindInterfaceClustersInUserGroup(groupArchString, userGroupSeqId, pkinaseEntries);
}
/// <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);*/
}
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>
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 AddSymmetryIndexes()
{
DbBuilderHelper.Initialize();
ChainInterfaces.InterfaceSymmetry interfaceSym = new ChainInterfaces.InterfaceSymmetry();
interfaceSym.CalculateInterfaceSymmetryJindex();
}
/// <summary>
/// build from current database by adding update/new pdb entries
/// but rebuild entry-based groups due to new PFAM version
/// regroup crystal forms
/// </summary>
public void BuildEntryInterfaceGroups(int step)
{
DbBuilderHelper.Initialize();
ProtCidSettings.progressInfo.Reset();
ProtCidSettings.progressInfo.totalOperationNum = 8;
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Building entry-based groups, includes 8 steps: ");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("1.Calculate entry interfaces");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("2. Generate interface files, calculate ASA.");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("3. Compare cryst interfaces with PDB/PQS/PISA BU and ASU Interfaces");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("4. Classify PDB entries into homologous groups");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("5. Retrieve chain alignments in groups");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("6. Calculate similarity Q scores in an entry group.");
// ProtCidSettings.progressInfo.progStrQueue.Enqueue("7. Compare interfaces of entries in entry groups.");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("7. Deal With Redundant Crystal Forms.");
ProtCidSettings.logWriter.WriteLine(DateTime.Today.ToShortTimeString());
ProtCidSettings.logWriter.WriteLine("Build entry-based group data.");
// update entry-related data, it is not necessary to recompute entry data from scratch
string[] updateEntries = GetMissBuCompEntries();
CrystInterfaceBuilder interfaceBuilder = new CrystInterfaceBuilder();
switch (step)
{
case 1:
ProtCidSettings.logWriter.WriteLine("1. Compute interfaces from crystals");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("1. Compute interfaces from crystals");
interfaceBuilder.ComputeEntryCrystInterfaces(updateEntries);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
goto case 2;
case 2:
ProtCidSettings.logWriter.WriteLine("2. Generate interface files, calculate ASA and contacts in interfaces");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("2. Generate interface files, calculate ASA and contacts in interfaces");
interfaceBuilder.GenerateEntryInterfaceFiles(updateEntries);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
break;
case 3:
ProtCidSettings.logWriter.WriteLine("3. Compare cryst interfaces with PDB/PISA BU and ASU Interfaces");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("3. Compare cryst interfaces with PDB/PISA BU and ASU Interfaces");
// compare cryst interfaces and BU interfaces for each entry
EntryCrystBuInterfaceComp crystBuInterfaceComp = new EntryCrystBuInterfaceComp();
crystBuInterfaceComp.UpdateEntryCrystBuInterfaces(updateEntries);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("CrystBuInterfaceComp", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
break;
case 4:
// group entries, rebuild group tables: PfamHomoSeqInfo, PfamHomoGroupEntryAlign, PfamHomoRepEntryAlign
ProtCidSettings.logWriter.WriteLine("4. Classify PDB entries into homologous groups.");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("4. Classify PDB entries into homologous groups");
PfamEntryClassifier pfamClassifier = new PfamEntryClassifier();
pfamClassifier.ClassifyPdbPfamGroups();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("PFAMHOMO", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
break;
case 5:
// retrieve chain alignments from HH, and fatcat alignments
ProtCidSettings.logWriter.WriteLine("5. Retrieve chain alignments in groups");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("5. Retrieve chain alignments in groups");
Alignments.GroupEntryAlignments groupAlignment = new InterfaceClusterLib.Alignments.GroupEntryAlignments();
groupAlignment.UpdateCrcHhAlignments();
groupAlignment.RetrieveChainAlignmentsInGroups();
// groupAlignment.GetRepChainPairsToBeAligned();
// groupAlignment.GetRepEntryChainAlignFilesFromMissingEntityAlign();
// groupAlignment.UpdateFatcatAlignmentsFromFile("");
// int[] updateGroups = { 10400 };
// groupAlignment.UpdateGroupEntryAlignments (updateGroups);
// groupAlignment.UpdateMissingChainAlignmentsInGroups (null);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("PFAMHOMO", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
break;
case 6:
// compare entry interfaces in entry groups
ProtCidSettings.logWriter.WriteLine("6. Calculate similarity Q scores in a entry-grouped group");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("6. Calculate similarity Q scores in a entry-grouped group");
HomoGroupInterfacesFinder.modifyType = "build";
HomoGroupInterfacesFinder groupInterfaceFinder = new HomoGroupInterfacesFinder();
groupInterfaceFinder.DetectHomoGroupInterfaces(null);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("INTERFACE", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
break;
case 7:
// group crystal forms based on Q scores of entry interfaces in an entry group
ProtCidSettings.logWriter.WriteLine("7. Detecting redundant crystal forms");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("6. Detecting redundant crystal forms");
RedundantCrystForms reduntCrystForm = new RedundantCrystForms();
reduntCrystForm.CheckReduntCrystForms();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
// distinct crystal forms
ProtCidSettings.logWriter.WriteLine("8. Cluster crystal forms in entry groups");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Non-Redundant Crystal Forms");
NonredundantCfGroups nonreduntCfGroups = new NonredundantCfGroups();
nonreduntCfGroups.UpdateCfGroupInfo();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("REDUN", ProtCidSettings.protcidDbConnection);
break;
default:
break;
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Build entry-based groups Done!");
ProtCidSettings.logWriter.WriteLine("Build entry-based groups Done!");
ProtCidSettings.logWriter.Flush();
}
/// <summary>
/// update database
/// </summary>
public void UpdatePfamInterfaces()
{
DbBuilderHelper.Initialize();
ProtCidSettings.progressInfo.Reset();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Updating PFAM Interfaces Data.");
string line = "";
string[] updateEntries = null;
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Get the entries to be updated from XML file directory.");
// string lsFile = "EntriesNotInEntryLevelGroups.txt";
updateEntries = GetUpdateEntries();
// updateEntries = GetUpdateEntries(lsFile);
if (updateEntries.Length == 0)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue("No update needed.");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.threadFinished = true;
return;
}
// Update entry interfaces,
// generate interface files, calculate surface area,
// then compare crystal interfaces to interfaces in PDB/PISA BAs
// UpdateEntryInterfaceData(updateEntries);
// updateEntries = GetUpdateEntriesFromPfamUpdate();
// classify these groups
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Update classification of homologous groups");
PfamEntryClassifier pfamClassifier = new PfamEntryClassifier();
Dictionary <int, string[]> updateGroupHash = null;
updateGroupHash = pfamClassifier.UpdatePdbPfamGroups(updateEntries);
ProtCidSettings.progressInfo.currentOperationIndex++;
if (updateGroupHash == null)
{
updateGroupHash = new Dictionary <int, string[]>();
}
// divide into two groups
/* int groupIdDivide = 7000;
* ProtCidSettings.progressInfo.progStrQueue.Enqueue("This is for groupId >= 7000");
* ProtCidSettings.logWriter.WriteLine("This is for groupid >= 7000.");*/
if (updateGroupHash.Count == 0)
{
int groupId = 0;
StreamReader updateGroupReader = new StreamReader("updateGroups.txt");
// string line = "";
while ((line = updateGroupReader.ReadLine()) != null)
{
string[] fields = line.Split(' ');
groupId = Convert.ToInt32(fields[0]);
// if (groupId >= groupIdDivide)
string[] entries = new string[fields.Length - 1];
Array.Copy(fields, 1, entries, 0, entries.Length);
updateGroupHash.Add(groupId, entries);
}
updateGroupReader.Close();
}
int[] updateGroups = new int[updateGroupHash.Count];
updateGroupHash.Keys.CopyTo(updateGroups, 0);
Array.Sort(updateGroups);
// update xtal interfaces for these groups
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Retrieve alignments in a group");
Alignments.GroupEntryAlignments groupAlignment = new InterfaceClusterLib.Alignments.GroupEntryAlignments();
try
{
groupAlignment.UpdateChainAlignmentsInGroups(updateGroups);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(ex.Message);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Stop the updating alignments and return.");
ProtCidSettings.logWriter.WriteLine(ex.Message);
ProtCidSettings.logWriter.WriteLine("Stop the updating alignments and return.");
ProtCidSettings.logWriter.Flush();
return;
}
ProtCidSettings.progressInfo.currentOperationIndex++;
// update xtal interfaces for these groups
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Build crystal, detect common interfaces in a group");
HomoGroupInterfacesFinder.modifyType = "update";
HomoGroupInterfacesFinder groupInterfaceFinder = new HomoGroupInterfacesFinder();
try
{
groupInterfaceFinder.UpdateHomoGroupInterfaces(updateGroupHash);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(ex.Message);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Stop the updating group interfaces comparing and return.");
ProtCidSettings.logWriter.WriteLine(ex.Message);
ProtCidSettings.logWriter.WriteLine("Stop the updating group interfaces comparing and return.");
ProtCidSettings.logWriter.Flush();
return;
}
ProtCidSettings.progressInfo.currentOperationIndex++;
ProtCidSettings.progressInfo.progStrQueue.Enqueue("6. Detecting redundant crystal forms");
RedundantCrystForms reduntCrystForm = new RedundantCrystForms();
try
{
reduntCrystForm.UpdateReduntCrystForms(updateGroups);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(ex.Message);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Stop the updating redundant cryst forms and return.");
ProtCidSettings.logWriter.WriteLine(ex.Message);
ProtCidSettings.logWriter.WriteLine("Stop the updating redundant cryst forms and return.");
ProtCidSettings.logWriter.Flush();
return;
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.currentOperationIndex++;
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Non-Redundant Crystal Forms");
NonredundantCfGroups nonreduntCfGroups = new NonredundantCfGroups();
try
{
nonreduntCfGroups.UpdateCfGroupInfo(updateGroups);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(ex.Message);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Stop the updating CF groups and return.");
ProtCidSettings.logWriter.WriteLine(ex.Message);
ProtCidSettings.logWriter.WriteLine("Stop the updating CF groups and return.");
ProtCidSettings.logWriter.Flush();
return;
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Recomputing indices.");
DbBuilderHelper.UpdateIndexes("RedundantCfGroups", ProtCidSettings.protcidDbConnection);
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
ProtCidSettings.logWriter.Close();
ProtCidSettings.progressInfo.threadFinished = true;
}
