/// <summary>
///
/// </summary>
public void CompressBigRelations()
{
int[] relSeqIds = { 2055, 2124, 2178, 11619, 11642, 14647, 14811 };
ProtCidSettings.dirSettings.seqFastaPath = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\fasta", "\\DomainFasta");
webFastaFileDir = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\DomainFasta", "\\WebDomainFasta");
List <string> seqFastaFileList = new List <string>();
string relationSeqFile = "";
string srcRelSeqFile = "";
string destRelSeqFile = "";
// string relationFolder = "";
string relationName = "";
foreach (int relSeqId in relSeqIds)
{
seqFastaFileList.Clear();
string[] relFastaFiles = Directory.GetFiles(ProtCidSettings.dirSettings.seqFastaPath, "Cluster" + relSeqId.ToString() + "A_*.fasta");
foreach (string fastaFile in relFastaFiles)
{
FileInfo fileInfo = new FileInfo(fastaFile);
seqFastaFileList.Add(fileInfo.Name);
}
relFastaFiles = Directory.GetFiles(ProtCidSettings.dirSettings.seqFastaPath, "Cluster" + relSeqId.ToString() + "B_*.fasta");
foreach (string fastaFile in relFastaFiles)
{
FileInfo fileInfo = new FileInfo(fastaFile);
seqFastaFileList.Add(fileInfo.Name);
}
seqFastaFileList.Add("Group" + relSeqId.ToString() + "A.fasta");
if (File.Exists(Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, "Group" + relSeqId.ToString() + "B.fasta")))
{
seqFastaFileList.Add("Group" + relSeqId.ToString() + "B.fasta");
}
string[] seqFastaFiles = new string[seqFastaFileList.Count];
seqFastaFileList.CopyTo(seqFastaFiles);
// relationFolder = MoveSeqFastaFilesToGroupFolder (seqFastaFiles, ProtCidSettings.dirSettings.seqFastaPath, relSeqId);
// relationSeqFile = TarFastaFilesOnFolder (relSeqId, relationFolder);
relationName = DownloadableFileName.GetDomainRelationName(relSeqId);
string seqTarFile = "Seq_" + relationName + ".tar.gz";
fileCompress.RunTar(seqTarFile, seqFastaFiles, ProtCidSettings.dirSettings.seqFastaPath, true);
// move the tar file to the web folder
srcRelSeqFile = Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, relationSeqFile);
destRelSeqFile = Path.Combine(webFastaFileDir, relationSeqFile);
if (File.Exists(destRelSeqFile))
{
File.Delete(destRelSeqFile);
}
File.Move(srcRelSeqFile, destRelSeqFile);
}
}
/// <summary>
///
/// </summary>
public string GenerateDomainGroupIdFileNameMatch()
{
string domainGroupIdNameLsFile = Path.Combine(ProtCidSettings.dirSettings.interfaceFilePath, "DomainGroupIdNameMap.txt");
StreamWriter lsFileWriter = new StreamWriter(domainGroupIdNameLsFile);
string queryString = "Select Distinct RelSeqID From PfamDomainFamilyRelation;";
DataTable domainGroupTable = ProtCidSettings.protcidQuery.Query(queryString);
int relSeqId = 0;
string groupName = "";
foreach (DataRow domainGroupRow in domainGroupTable.Rows)
{
relSeqId = Convert.ToInt32(domainGroupRow["RelSeqID"].ToString());
groupName = DownloadableFileName.GetDomainRelationName(relSeqId);
lsFileWriter.Write(relSeqId + "\t" + groupName + "\n");
}
lsFileWriter.Close();
return(domainGroupIdNameLsFile);
}
/// <summary>
///
/// </summary>
public string GenerateChainGroupIdFileNameMatch()
{
string chainGroupIdNameLsFile = Path.Combine(ProtCidSettings.dirSettings.interfaceFilePath, "ChainGroupIdNameMap.txt");
StreamWriter lsFileWriter = new StreamWriter(chainGroupIdNameLsFile);
string queryString = "Select Distinct SuperGroupSeqID From PfamSuperGroups;";
DataTable chainGroupTable = ProtCidSettings.protcidQuery.Query(queryString);
int chainGroupId = 0;
string groupName = "";
foreach (DataRow chainGroupRow in chainGroupTable.Rows)
{
chainGroupId = Convert.ToInt32(chainGroupRow["SuperGroupSeqID"].ToString());
groupName = DownloadableFileName.GetChainGroupTarGzFileName(chainGroupId);
lsFileWriter.Write(chainGroupId + "\t" + groupName + "\n");
}
lsFileWriter.Close();
return(chainGroupIdNameLsFile);
}
/// <summary>
///
/// </summary>
/// <param name="superGroupId"></param>
/// <param name="clusterReverseFilesHash"></param>
public void CompressGroupClustersFiles(int superGroupId)
{
ProtCidSettings.progressInfo.currentOperationNum++;
ProtCidSettings.progressInfo.currentStepNum++;
ProtCidSettings.progressInfo.currentFileName = superGroupId.ToString();
int[] groupClusters = GetClustersForGroup(superGroupId);
string groupName = DownloadableFileName.GetChainGroupTarGzFileName(superGroupId);
List <string> clusterFileList = new List <string> ();
string clusterFile = "";
foreach (int clusterId in groupClusters)
{
ProtCidSettings.progressInfo.currentFileName = superGroupId.ToString() + "_" + clusterId.ToString();
try
{
clusterFile = CompressGroupClusterInterfaceFiles(superGroupId, clusterId, groupName);
if (clusterFile != "")
{
clusterFileList.Add(clusterFile);
}
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(superGroupId.ToString() + "_" + clusterId.ToString() +
"Compress cluster interface files errors: " + ex.Message);
ProtCidSettings.logWriter.WriteLine(superGroupId.ToString() + "_" + clusterId.ToString() +
"Compress cluster interface files errors: " + ex.Message);
ProtCidSettings.logWriter.Flush();
}
}
// tar cluster files to group
string groupTarFileName = groupName + ".tar";
fileCompress.RunTar(groupTarFileName, clusterFileList.ToArray(), clusterFileDir, false);
}
/// <summary>
///
/// </summary>
/// <param name="superGroupIds"></param>
public void UpdateClusterFastaFiles(int[] superGroupIds)
{
webFastaFileDir = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\ChainFasta", "\\webChainFasta");
if (!Directory.Exists(webFastaFileDir))
{
Directory.CreateDirectory(webFastaFileDir);
}
// string[] allGroupFastaFiles = Directory.GetFiles(ProtCidSettings.dirSettings.seqFastaPath, "group*.fasta");
string groupIdWithType = "";
string clusterInfoFile = "";
List <string> fastaFileList = new List <string> ();
string tarFile = "";
ProtCidSettings.progressInfo.ResetCurrentProgressInfo();
ProtCidSettings.progressInfo.totalOperationNum = superGroupIds.Length;
ProtCidSettings.progressInfo.totalStepNum = superGroupIds.Length;
ProtCidSettings.progressInfo.currentOperationLabel = "Tar Seq Files";
foreach (int groupId in superGroupIds)
{
ProtCidSettings.progressInfo.currentFileName = groupId.ToString();
ProtCidSettings.progressInfo.currentStepNum++;
ProtCidSettings.progressInfo.currentOperationNum++;
fastaFileList.Clear();
string[] groupFastaFiles = GetGroupSeqFastaFiles(groupId);
foreach (string groupFastaFile in groupFastaFiles)
{
groupIdWithType = GetGroupIdWithTypeFromFileName(groupFastaFile);
fastaFileList.Add("Group" + groupIdWithType + ".fasta");
try
{
clusterInfoFile = Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, "Cluster" + groupIdWithType + ".txt");
Dictionary <string, string> groupEntitySequenceHash = null;
Dictionary <string, string> groupEntityAnnotationHash = null;
Dictionary <int, string[]> clusterEntityHash = GetClusterEntityHash(clusterInfoFile);
ReadEntitySequenceHash(groupFastaFile, out groupEntitySequenceHash, out groupEntityAnnotationHash);
foreach (int clusterId in clusterEntityHash.Keys)
{
string[] clusterEntities = (string[])clusterEntityHash[clusterId];
string clusterFastaFile = WriteClusterEntitySequencesToFile(groupIdWithType, clusterId, clusterEntities, groupEntitySequenceHash, groupEntityAnnotationHash);
fastaFileList.Add("Cluster" + groupIdWithType + "_" + clusterId.ToString() + ".fasta");
}
}
catch (Exception ex)
{
logWriter.WriteLine("Output cluster fasta files errors: " + groupFastaFile + " " + ex.Message);
logWriter.Flush();
}
}
string[] fastaFiles = new string[fastaFileList.Count];
fastaFileList.CopyTo(fastaFiles);
string chainGroupName = DownloadableFileName.GetChainGroupTarGzFileName(groupId);
try
{
DeleteObsoleteWebFastaFiles(chainGroupName);
// tarFile = TarFastaFiles(groupId, fastaFiles);
tarFile = "Seq_" + chainGroupName + ".tar.gz";
tarFile = fileCompress.RunTar(tarFile, fastaFiles, ProtCidSettings.dirSettings.seqFastaPath, true);
File.Move(Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, tarFile), Path.Combine(webFastaFileDir, tarFile));
}
catch (Exception ex)
{
logWriter.WriteLine("Tar and move file error for " + tarFile + " : " + ex.Message);
logWriter.Flush();
}
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Tar sequence files done!");
ProtCidSettings.logWriter.WriteLine("Tar sequence files done!");
}
/// <summary>
///
/// </summary>
public void PrintClusterFastaFiles()
{
string[] allGroupFastaFiles = Directory.GetFiles(ProtCidSettings.dirSettings.seqFastaPath, "group*.fasta");
Dictionary <int, List <string> > groupFastaFilesHash = GetGroupIDsFromFiles(allGroupFastaFiles);
string groupIdWithType = "";
string clusterInfoFile = "";
List <string> fastaFileList = new List <string> ();
List <int> groupIdList = new List <int> (groupFastaFilesHash.Keys);
groupIdList.Sort();
string tarFile = "";
ProtCidSettings.progressInfo.ResetCurrentProgressInfo();
ProtCidSettings.progressInfo.currentOperationLabel = "Tar Seq Files";
ProtCidSettings.progressInfo.totalOperationNum = groupIdList.Count;
ProtCidSettings.progressInfo.totalStepNum = groupIdList.Count;
foreach (int chainGroupId in groupIdList)
{
ProtCidSettings.progressInfo.currentFileName = chainGroupId.ToString();
ProtCidSettings.progressInfo.currentOperationNum++;
ProtCidSettings.progressInfo.currentStepNum++;
fastaFileList.Clear();
foreach (string groupFastaFile in groupFastaFilesHash[chainGroupId])
{
groupIdWithType = GetGroupIdWithTypeFromFileName(groupFastaFile);
fastaFileList.Add("Group" + groupIdWithType + ".fasta");
try
{
clusterInfoFile = Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, "Cluster" + groupIdWithType + ".txt");
Dictionary <string, string> groupEntitySequenceHash = null;
Dictionary <string, string> groupEntityAnnotationHash = null;
Dictionary <int, string[]> clusterEntityHash = GetClusterEntityHash(clusterInfoFile);
ReadEntitySequenceHash(groupFastaFile, out groupEntitySequenceHash, out groupEntityAnnotationHash);
foreach (int clusterId in clusterEntityHash.Keys)
{
string[] clusterEntities = (string[])clusterEntityHash[clusterId];
string clusterFastaFile = WriteClusterEntitySequencesToFile(groupIdWithType, clusterId, clusterEntities, groupEntitySequenceHash, groupEntityAnnotationHash);
fastaFileList.Add("Cluster" + groupIdWithType + "_" + clusterId.ToString() + ".fasta");
}
}
catch (Exception ex)
{
logWriter.WriteLine("Output cluster fasta files errors: " + groupFastaFile + " " + ex.Message);
logWriter.Flush();
}
}
string[] fastaFiles = new string[fastaFileList.Count];
fastaFileList.CopyTo(fastaFiles);
string chainGroupName = DownloadableFileName.GetChainGroupTarGzFileName(chainGroupId);
try
{
string fastaTarFile = "Seq_" + chainGroupName + ".tar.gz";
/* if (fastaFiles.Length > 100)
* {
* string groupFolder = MoveSeqFastaFilesToGroupFolder(fastaFiles, ProtCidSettings.dirSettings.seqFastaPath, groupId);
* tarFile = TarFastaFilesOnFolder(groupId, groupFolder);
* }
* else
* {
* tarFile = TarFastaFiles(groupId, fastaFiles);
* }*/
fastaTarFile = fileCompress.RunTar(fastaTarFile, fastaFiles, ProtCidSettings.dirSettings.seqFastaPath, true);
File.Move(Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, fastaTarFile), Path.Combine(webFastaFileDir, fastaTarFile));
}
catch (Exception ex)
{
logWriter.WriteLine("Tar and move file error for " + tarFile + " : " + ex.Message);
logWriter.Flush();
}
}
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Tar sequence files done!");
}
/// <summary>
///
/// </summary>
public void UpdateClusterDomainSequencesForDebug(int[] relSeqIds)
{
ProtCidSettings.dirSettings.seqFastaPath = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\fasta", "\\DomainFasta");
if (!Directory.Exists(ProtCidSettings.dirSettings.seqFastaPath))
{
Directory.CreateDirectory(ProtCidSettings.dirSettings.seqFastaPath);
}
webFastaFileDir = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\DomainFasta", "\\WebDomainFasta");
if (!Directory.Exists(webFastaFileDir))
{
Directory.CreateDirectory(webFastaFileDir);
}
StreamWriter lsFileWriter = new StreamWriter(Path.Combine(webFastaFileDir, "relSeq-newls.txt"), true);
string relationSeqFile = "";
ProtCidSettings.progressInfo.ResetCurrentProgressInfo();
ProtCidSettings.progressInfo.totalOperationNum = relSeqIds.Length;
ProtCidSettings.progressInfo.totalStepNum = relSeqIds.Length;
string srcRelSeqFile = "";
string destRelSeqFile = "";
string relationName = "";
foreach (int relSeqId in relSeqIds)
{
ProtCidSettings.progressInfo.currentFileName = relSeqId.ToString();
ProtCidSettings.progressInfo.currentOperationNum++;
ProtCidSettings.progressInfo.currentStepNum++;
try
{
string[] fastaSeqFiles = GetRelationClusterSeqFastaFiles(relSeqId);
relationName = DownloadableFileName.GetDomainRelationName(relSeqId);
relationSeqFile = "Seq_" + relationName + ".tar.gz";
fileCompress.RunTar(relationSeqFile, fastaSeqFiles, ProtCidSettings.dirSettings.seqFastaPath, true);
lsFileWriter.WriteLine(relationSeqFile);
lsFileWriter.Flush();
// move the tar file to the web folder
srcRelSeqFile = Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, relationSeqFile);
destRelSeqFile = Path.Combine(webFastaFileDir, relationSeqFile);
if (File.Exists(destRelSeqFile))
{
File.Delete(destRelSeqFile);
}
File.Move(srcRelSeqFile, destRelSeqFile);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(relSeqId.ToString() + " Writing sequences to fasta files errors: " + ex.Message);
ProtCidSettings.logWriter.WriteLine(relSeqId.ToString() + " Writing sequences to fasta files errors: " + ex.Message);
ProtCidSettings.logWriter.Flush();
}
}
lsFileWriter.Close();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
}
/// <summary>
///
/// </summary>
public void PrintClusterDomainSequences()
{
ProtCidSettings.dirSettings.seqFastaPath = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\fasta", "\\DomainFasta");
if (!Directory.Exists(ProtCidSettings.dirSettings.seqFastaPath))
{
Directory.CreateDirectory(ProtCidSettings.dirSettings.seqFastaPath);
}
webFastaFileDir = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\DomainFasta", "\\WebDomainFasta");
if (!Directory.Exists(webFastaFileDir))
{
Directory.CreateDirectory(webFastaFileDir);
}
StreamWriter lsFileWriter = new StreamWriter(Path.Combine(webFastaFileDir, "relSeq-ls.txt"), true);
string relationSeqFile = "";
string queryString = "Select Distinct RelSeqID From PfamDomainInterfaceCluster;";
DataTable relSeqIdTable = ProtCidSettings.protcidQuery.Query(queryString);
int relSeqId = 0;
ProtCidSettings.progressInfo.ResetCurrentProgressInfo();
ProtCidSettings.progressInfo.totalOperationNum = relSeqIdTable.Rows.Count;
ProtCidSettings.progressInfo.totalStepNum = relSeqIdTable.Rows.Count;
queryString = "Select PdbID, EntityID, AsymID, Sequence From AsymUnit WHere PolymerType = 'polypeptide';";
DataTable entitySeqTable = ProtCidSettings.pdbfamQuery.Query(queryString);
queryString = "Select PdbID, DomainID, EntityID, SeqStart, SeqEnd, Pfam_ID, Pfam_Acc From PdbPfam;";
DataTable domainTable = ProtCidSettings.pdbfamQuery.Query(queryString);
string srcRelSeqFile = "";
string destRelSeqFile = "";
string relationName = "";
foreach (DataRow relSeqIdRow in relSeqIdTable.Rows)
{
relSeqId = Convert.ToInt32(relSeqIdRow["RelSeqID"].ToString());
relationName = DownloadableFileName.GetDomainRelationName(relSeqId);
ProtCidSettings.progressInfo.currentFileName = relSeqId.ToString();
ProtCidSettings.progressInfo.currentOperationNum++;
ProtCidSettings.progressInfo.currentStepNum++;
try
{
relationSeqFile = "Seq_" + relationName + ".tar.gz";
if (!File.Exists(relationSeqFile))
{
string[] fastaSeqFiles = PrintRelationClusterSeqFasta(relSeqId, entitySeqTable, domainTable);
relationSeqFile = fileCompress.RunTar(relationSeqFile, fastaSeqFiles, ProtCidSettings.dirSettings.seqFastaPath, true);
}
lsFileWriter.WriteLine(relationSeqFile);
lsFileWriter.Flush();
// move the tar file to the web folder
srcRelSeqFile = Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, relationSeqFile);
destRelSeqFile = Path.Combine(webFastaFileDir, relationSeqFile);
if (File.Exists(destRelSeqFile))
{
File.Delete(destRelSeqFile);
}
File.Move(srcRelSeqFile, destRelSeqFile);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(relSeqId.ToString() + " Writing sequences to fasta files errors: " + ex.Message);
ProtCidSettings.logWriter.WriteLine(relSeqId.ToString() + " Writing sequences to fasta files errors: " + ex.Message);
ProtCidSettings.logWriter.Flush();
}
}
lsFileWriter.Close();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
}
/// <summary>
///
/// </summary>
public void UpdateClusterDomainSequences(int[] relSeqIds)
{
ProtCidSettings.dirSettings.seqFastaPath = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\fasta", "\\DomainFasta");
if (!Directory.Exists(ProtCidSettings.dirSettings.seqFastaPath))
{
Directory.CreateDirectory(ProtCidSettings.dirSettings.seqFastaPath);
}
webFastaFileDir = ProtCidSettings.dirSettings.seqFastaPath.Replace("\\DomainFasta", "\\WebDomainFasta");
if (Directory.Exists(webFastaFileDir))
{
Directory.Delete(webFastaFileDir, true);
}
Directory.CreateDirectory(webFastaFileDir);
StreamWriter lsFileWriter = new StreamWriter(Path.Combine(webFastaFileDir, "relSeq-newls.txt"), true);
string relationSeqFile = "";
ProtCidSettings.progressInfo.ResetCurrentProgressInfo();
ProtCidSettings.progressInfo.totalOperationNum = relSeqIds.Length;
ProtCidSettings.progressInfo.totalStepNum = relSeqIds.Length;
string queryString = "Select PdbID, EntityID, AsymID, Sequence From AsymUnit WHere PolymerType = 'polypeptide';";
DataTable entitySeqTable = ProtCidSettings.pdbfamQuery.Query(queryString);
queryString = "Select PdbID, DomainID, EntityID, SeqStart, SeqEnd, Pfam_ID, Pfam_Acc From PdbPfam;";
DataTable domainTable = ProtCidSettings.pdbfamQuery.Query(queryString);
string srcRelSeqFile = "";
string destRelSeqFile = "";
string relationName = "";
foreach (int relSeqId in relSeqIds)
{
ProtCidSettings.progressInfo.currentFileName = relSeqId.ToString();
ProtCidSettings.progressInfo.currentOperationNum++;
ProtCidSettings.progressInfo.currentStepNum++;
try
{
DeleteRelationSeqFiles(relSeqId); // delete the old files
string[] fastaSeqFiles = PrintRelationClusterSeqFasta(relSeqId, entitySeqTable, domainTable);
// the length of parameters of Cmd.exe cannot be longer than 8191
relationName = DownloadableFileName.GetDomainRelationName(relSeqId);
relationSeqFile = "Seq_" + relationName + ".tar.gz";
fileCompress.RunTar(relationSeqFile, fastaSeqFiles, ProtCidSettings.dirSettings.seqFastaPath, true);
lsFileWriter.WriteLine(relationSeqFile);
lsFileWriter.Flush();
// move the tar file to the web folder
srcRelSeqFile = Path.Combine(ProtCidSettings.dirSettings.seqFastaPath, relationSeqFile);
destRelSeqFile = Path.Combine(webFastaFileDir, relationSeqFile);
if (File.Exists(destRelSeqFile))
{
File.Delete(destRelSeqFile);
}
File.Move(srcRelSeqFile, destRelSeqFile);
}
catch (Exception ex)
{
ProtCidSettings.progressInfo.progStrQueue.Enqueue(relSeqId.ToString() + " Writing sequences to fasta files errors: " + ex.Message);
ProtCidSettings.logWriter.WriteLine(relSeqId.ToString() + " Writing sequences to fasta files errors: " + ex.Message);
ProtCidSettings.logWriter.Flush();
}
}
lsFileWriter.Close();
ProtCidSettings.progressInfo.progStrQueue.Enqueue("Done!");
}
