/// <summary>
///
/// </summary>
/// <param name="pfamId"></param>
private void UpdatePfamLigandsClustersHmm(string pfamId, List <string> updateEntryList)
{
DataTable jscoreTable = GetPfamComHmmTable(pfamId);
List <string> pdbDomainLigandList = GetPfamChainDomainLigandList(pfamId);
double[,] jscoreMatrix = SetConnectMatrix(pdbDomainLigandList, jscoreTable);
List <List <int> > existClusterList = GetExistingClusterList(pfamId, pdbDomainLigandList);
int[] updateLigandIndexes = GetUpdateLigandIndexes(updateEntryList, pdbDomainLigandList);
List <int[]> indexClusterList = ClusterPfamLigandsIndexes(jscoreMatrix, updateLigandIndexes, existClusterList);
int clusterId = 1;
string[] domainLigandFields = null;
foreach (int[] cluster in indexClusterList)
{
foreach (int ligandIndex in cluster)
{
domainLigandFields = pdbDomainLigandList[ligandIndex].Split('_');
DataRow dataRow = pfamLigandClusterTable.NewRow();
dataRow["PfamID"] = pfamId;
dataRow["ClusterID"] = clusterId;
dataRow["PdbID"] = domainLigandFields[0];
dataRow["ChainDomainID"] = domainLigandFields[1];
dataRow["LigandChain"] = domainLigandFields[2];
pfamLigandClusterTable.Rows.Add(dataRow);
}
clusterId++;
}
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.protcidDbConnection, pfamLigandClusterTable);
pfamLigandClusterTable.Clear();
}
/// <summary>
///
/// </summary>
/// <param name="ligandComAtomTable"></param>
public void ClusterPfamLigands(DataTable ligandComAtomTable)
{
// string[] clusterCols = { "PfamID", "ClusterID", "PdbID", "ChainDomainID", "LigandChain" };
List <string> pdbLigandList = GetPfamLigandsInOrder(ligandComAtomTable);
int[,] connectMatrix = SetConnectMatrix(pdbLigandList, ligandComAtomTable);
List <List <int> > clusterList = ClusterPfamLigands(connectMatrix);
List <List <int> > sortedClusterList = SortClustersInLigandNumbers(clusterList);
int clusterId = 1;
string[] ligandFields = null;
string pfamId = ligandComAtomTable.Rows[0]["PfamID"].ToString();
foreach (List <int> cluster in sortedClusterList)
{
foreach (int ligandIndex in cluster)
{
ligandFields = pdbLigandList[ligandIndex].Split('_');
DataRow clusterRow = pfamLigandClusterTable.NewRow();
clusterRow["PfamID"] = pfamId;
clusterRow["ClusterID"] = clusterId;
clusterRow["PdbID"] = ligandFields[0];
clusterRow["ChainDomainID"] = ligandFields[1];
clusterRow["LigandChain"] = ligandFields[2];
clusterRow["LigandFileChain"] = ligandFields[3];
pfamLigandClusterTable.Rows.Add(clusterRow);
}
clusterId++;
}
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.protcidDbConnection, pfamLigandClusterTable);
pfamLigandClusterTable.Clear();
}
/// <summary>
///
/// </summary>
/// <param name="updateEntries"></param>
private string[] ImportNewEntriesCrcCodes(string[] updateEntries)
{
int numOfEntries = 30;
string queryString = "Select * From PDBCRCMap;";
DataTable piscesPdbCrcTable = dbQuery.Query(remotePiscesDbConnect, queryString);
ArrayList updateCrcList = new ArrayList();
for (int i = 0; i < updateEntries.Length; i += numOfEntries)
{
if (i + numOfEntries >= updateEntries.Length)
{
numOfEntries = updateEntries.Length - i;
}
string[] queryEntries = new string[numOfEntries];
Array.Copy(updateEntries, i, queryEntries, 0, numOfEntries);
queryString = string.Format("Select * From PDBCRCMap Where pdb IN {0};", ParseHelper.FormatSqlListString(queryEntries));
DataTable pdbCrcTable = dbQuery.Query(remotePiscesDbConnect, queryString);
pdbCrcTable.TableName = "pdbcrcmap";
dbInsert.BatchInsertDataIntoDBtables(localMysqlDbConnect, pdbCrcTable);
string[] crcCodesInTable = GetCrcCodesInTable(pdbCrcTable);
foreach (string crc in crcCodesInTable)
{
if (!updateCrcList.Contains(crc))
{
updateCrcList.Add(crc);
}
}
}
string[] updateCrcCodes = new string[updateCrcList.Count];
updateCrcList.CopyTo(updateCrcCodes);
return(updateCrcCodes);
}
/// <summary>
/// build hh alignment table
/// </summary>
/// <param name="hhrListFile">the list of sequence codes, like uniprot code, pdb id, crc code</param>
/// <param name="hhrFileDir">the directory where hhr files are</param>
public void BuildHhAlignments(string hhrListFile, string hhrFileDir)
{
if (hhrListFile == "")
{
BuildHhAlignments(hhrFileDir);
return;
}
bool isUpdate = false;
Initialize(isUpdate);
StreamReader lsReader = new StreamReader(hhrListFile);
string seqCode = "";
string hhrFile = "";
bool deleteHhrFile = false;
while ((seqCode = lsReader.ReadLine()) != null)
{
deleteHhrFile = false;
hhrFile = Path.Combine(hhrFileDir, seqCode + ".hhr");
if (!File.Exists(hhrFile))
{
hhrFile = Path.Combine(hhrFileDir, seqCode + ".hhr.gz");
hhrFile = ParseHelper.UnZipFile(hhrFile, ProtCidSettings.tempDir);
deleteHhrFile = true;
}
try
{
ParseHhrFile(hhrFile, hhAlignTable);
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.alignmentDbConnection, hhAlignTable);
// InsertHhAlignDataToDb(hhAlignTable);
}
catch (Exception ex)
{
ProtCidSettings.logWriter.WriteLine(seqCode + " Parsing hhr file error: " + ex.Message);
ProtCidSettings.logWriter.Flush();
}
hhAlignTable.Clear();
if (deleteHhrFile)
{
File.Delete(hhrFile);
}
}
lsReader.Close();
ProtCidSettings.logWriter.Flush();
}
/* Modified on Feb. 02, 2009
* The alignments will be added at the next step,
* because the best alignment is choosed from PSIBLAST, CE and FATCAT alignments.
* And missed alignments will be added by outside of the project (FATCAT/CE no Windows executable)
*/
/// <summary>
///
/// </summary>
/// <param name="sgEntryHash"></param>
public void AddGroupChainPairsToTables(Dictionary <string, List <string> > sgEntryHash)
{
List <string> repPdbList = new List <string> ();
#if DEBUG
string repEntryString = "";
string repHomoEntryString = "";
int subGroupNum = 1;
#endif
foreach (DataRow dRow in HomoGroupTables.homoGroupTables[HomoGroupTables.HomoSeqInfo].Rows)
{
List <string> pdbList = sgEntryHash[dRow["SpaceGroup"].ToString() + "_" + dRow["ASU"].ToString()];
pdbList.Remove(dRow["PdbID"].ToString());
#if DEBUG
repEntryString += dRow["PdbID"].ToString();
repEntryString += ",";
#endif
if (!repPdbList.Contains(dRow["PdbID"].ToString()))
{
repPdbList.Add(dRow["PdbID"].ToString());
}
if (pdbList.Count > 0)
{
AddDataToHomoRepEntryAlignTable(dRow["PdbID"].ToString(), pdbList, groupSeqNum);
// entryAlignment.RetrieveRepEntryAlignment(dRow["PdbID"].ToString(), pdbList, groupSeqNum);
#if DEBUG
repHomoEntryString = dRow["PdbID"].ToString();
foreach (string pdbId in pdbList)
{
repHomoEntryString += ",";
repHomoEntryString += pdbId;
}
subGroupNum++;
#endif
}
}
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.protcidDbConnection, HomoGroupTables.homoGroupTables[HomoGroupTables.HomoRepEntryAlign]);
HomoGroupTables.homoGroupTables[HomoGroupTables.HomoRepEntryAlign].Clear();
// retrieve pairwise entries information for each group
AddDataToRepEntryAlignTable(repPdbList, groupSeqNum);
// entryAlignment.GetEntryPairAlignment(repPdbList, groupSeqNum);
}
/// <summary>
///
/// </summary>
public void GenerateClansInPdbTable()
{
DbInsert dbInsert = new DbInsert();
DataTable pfamClansInPdbTable = CreateTable(false);
string queryString = "Select Distinct Clan_Acc From PfamInPdb, PfamClanFamily Where PfamInPdb.PfamAcc = PfamClanFamily.Pfam_Acc;";
DataTable clansTable = ProtCidSettings.pdbfamQuery.Query(queryString);
string clanAcc = "";
int numPfams = 0;
int numPfamsPdb = 0;
int numPfamsPeptide = 0;
int numPfamsDnaRna = 0;
int numEntries = 0;
int numUniProts = 0;
string[] clanPfamsPdb = null;
foreach (DataRow clanRow in clansTable.Rows)
{
clanAcc = clanRow["Clan_Acc"].ToString().TrimEnd();
DataRow clanInfoRow = pfamClansInPdbTable.NewRow();
clanInfoRow["ClanAcc"] = clanAcc;
numPfams = GetClanNumPfams(clanAcc);
numPfamsPdb = GetClanNumPfamsInPdb(clanAcc, out clanPfamsPdb);
numPfamsPeptide = GetNumPfamsWithPeptides(clanPfamsPdb);
numPfamsDnaRna = GetNumPfamsWithDnaRna(clanPfamsPdb);
numEntries = GetClanNumEntries(clanAcc);
numUniProts = GetClanNumUniProtsInPdb(clanAcc);
clanInfoRow["NumPfams"] = numPfams;
clanInfoRow["NumPfamsPdb"] = numPfamsPdb;
clanInfoRow["NumPfamsPeptide"] = numPfamsPeptide;
clanInfoRow["NumPfamsDnaRna"] = numPfamsDnaRna;
clanInfoRow["NumEntries"] = numEntries;
clanInfoRow["NumUniProts"] = numUniProts;
pfamClansInPdbTable.Rows.Add(clanInfoRow);
}
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.pdbfamDbConnection, pfamClansInPdbTable);
}
/// <summary>
/// parse one fatcat alignment output file
/// insert data into database
/// </summary>
/// <param name="alignFile"></param>
public void ParseFatcatAlignmentFile(string alignFile)
{
/* modified on April 5, 2010, change the input files of FATCAT from Guoli's regular files
* into my regular file with XMl sequential numbers and asymID
* instead of PDB sequence numbers.
* so no sequence nubmers conversion needed
*/
if (logWriter == null)
{
logWriter = new StreamWriter("fatcatAlignmentsLog.txt", true);
}
logWriter.WriteLine(alignFile);
StreamReader dataReader = new StreamReader(alignFile);
string line = "";
int scoreIdx = -1;
int alignLenIdx = -1;
int gapIdx = -1;
int gapEndIdx = -1;
string alignSequence1 = "";
string alignSequence2 = "";
int alignStart1 = -1;
int alignEnd1 = -1;
int alignStart2 = -1;
int alignEnd2 = -1;
string[] fields = null;
bool chain1Started = false;
bool chain2Started = false;
AlignSeqInfo alignInfo1 = new AlignSeqInfo();
AlignSeqInfo alignInfo2 = new AlignSeqInfo();
DataRow dataRow = null;
Dictionary <string, string> entryAuthChainHash = new Dictionary <string, string> ();
string dataLine = "";
while ((line = dataReader.ReadLine()) != null)
{
if (line == "")
{
continue;
}
if (line.IndexOf("Align") > -1 &&
line.Substring(0, "Align".Length) == "Align")
{
// the fatcat format: Align 3v2dX.pdb 70 with 4garZ.pdb 58
fields = ParseHelper.SplitPlus(line, ' ');
// get the pdbid and chain id from the fileName
string[] entryChainFields1 = GetEntryChainFields(fields[1]);
string[] entryChainFields2 = GetEntryChainFields(fields[4]);
dataRow = FatcatTables.fatcatAlignTable.NewRow();
dataRow["QueryEntry"] = entryChainFields1[0];
dataRow["QueryLength"] = fields[2];
dataRow["HitEntry"] = entryChainFields2[0];
dataRow["HitLength"] = fields[5];
alignInfo1.pdbId = entryChainFields1[0];
alignInfo1.asymChainId = entryChainFields1[1];
alignInfo1.chainId = GetAuthorChainFromAsymID(alignInfo1.pdbId, alignInfo1.asymChainId, ref entryAuthChainHash);
dataRow["QueryChain"] = alignInfo1.chainId;
dataRow["QueryAsymChain"] = alignInfo1.asymChainId;
alignInfo2.pdbId = entryChainFields2[0];
alignInfo2.asymChainId = entryChainFields2[1];
alignInfo2.chainId = GetAuthorChainFromAsymID(alignInfo2.pdbId, alignInfo2.asymChainId, ref entryAuthChainHash);
dataRow["HitAsymChain"] = alignInfo2.asymChainId;
dataRow["HitChain"] = alignInfo2.chainId;
alignSequence1 = "";
alignSequence2 = "";
chain1Started = true;
chain2Started = true;
dataLine = "";
}
dataLine += (line + "\r\n");
scoreIdx = line.IndexOf("Score");
if (scoreIdx > -1)
{
// from opt-equ, equivalent positions
alignLenIdx = line.IndexOf("align-len");
gapIdx = line.IndexOf("gaps");
gapEndIdx = line.LastIndexOf("(");
dataRow["Score"] = line.Substring(scoreIdx + "Score".Length + 1, alignLenIdx - scoreIdx - "Score".Length - 1);
dataRow["Align_Len"] = line.Substring(alignLenIdx + "align-len".Length + 1,
gapIdx - alignLenIdx - "align-len".Length - 2);
dataRow["Gaps"] = line.Substring(gapIdx + "gaps".Length + 1, gapEndIdx - gapIdx - "gaps".Length - 2);
}
if (line.IndexOf("P-value") > -1)
{
fields = ParseHelper.SplitPlus(line, ' ');
dataRow["E_Value"] = Convert.ToDouble(fields[1]);
dataRow["Identity"] = fields[5].TrimEnd('%');
dataRow["Similarity"] = fields[7].TrimEnd('%');
}
if (line.IndexOf("Chain 1: ") > -1)
{
fields = ParseHelper.SplitPlus(line, ' ');
if (chain1Started)
{
alignStart1 = ConvertSeqToInt(fields[2]);
chain1Started = false;
}
alignSequence1 += fields[3];
alignEnd1 = ConvertSeqToInt(fields[2]) + GetNonGapAlignedString(fields[3]).Length - 1;
}
if (line.IndexOf("Chain 2:") > -1)
{
line = line.Replace(':', ' ');
fields = ParseHelper.SplitPlus(line, ' ');
if (chain2Started)
{
alignStart2 = ConvertSeqToInt(fields[2]);
chain2Started = false;
}
alignSequence2 += fields[3];
alignEnd2 = ConvertSeqToInt(fields[2]) + GetNonGapAlignedString(fields[3]).Length - 1;
}
if (line.IndexOf("Note:") > -1)
{
alignInfo1.alignStart = alignStart1;
alignInfo1.alignEnd = alignEnd1;
alignInfo1.alignSequence = alignSequence1;
alignInfo2.alignStart = alignStart2;
alignInfo2.alignEnd = alignEnd2;
alignInfo2.alignSequence = alignSequence2;
/* if (IsAlignmentInDb(alignInfo1.pdbId, alignInfo1.chainId, alignInfo2.pdbId, alignInfo2.chainId))
* {
* continue;
* }
*/
// Convert aligned sequences to xml sequences
// add these residues with no-coordinate and no -Calpha to the alignment
// modified on August 31, 2012
try
{
seqConverter.AddDisorderResiduesToAlignment(ref alignInfo1, ref alignInfo2);
}
catch (Exception ex)
{
logWriter.WriteLine(alignInfo1.pdbId + alignInfo1.asymChainId + " " +
alignInfo2.pdbId + alignInfo2.asymChainId + " filling out disorder residues failed: " + ex.Message);
logWriter.Flush();
}
dataRow["AlignmentLength"] = GetAlignmentLength(alignSequence1, alignSequence2);
dataRow["QuerySequence"] = alignInfo1.alignSequence;
dataRow["HitSequence"] = alignInfo2.alignSequence;
// modified on April 10, 2010. Since input files for FATCAT use XML sequential numbers
dataRow["QueryStart"] = alignInfo1.alignStart;
dataRow["QueryEnd"] = alignInfo1.alignEnd;
dataRow["HitStart"] = alignInfo2.alignStart;
dataRow["HitEnd"] = alignInfo2.alignEnd;
// DeleteDbData(alignInfo1.pdbId, alignInfo1.chainId, alignInfo2.pdbId, alignInfo2.chainId);
FatcatTables.fatcatAlignTable.Rows.Add(dataRow);
/* try
* {
* dbInsert.InsertDataIntoDb(alignmentDbConnection, dataRow);
* }
* catch (Exception ex)
* {
* logWriter.WriteLine(alignFile + ": error " + ex.Message + "\r\n" + ParseHelper.FormatDataRow (dataRow) + " ");
* logWriter.Flush();
* }*/
}
}
dataReader.Close();
try
{
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.alignmentDbConnection, FatcatTables.fatcatAlignTable);
FatcatTables.fatcatAlignTable.Clear();
// "too many open handles to database", try to close the handles before leave by commit or rollback
dbUpdate.Update(ProtCidSettings.alignmentDbConnection, "Commit;");
}
catch (Exception ex)
{
logWriter.WriteLine(alignFile + ": error " + ex.Message + "\r\n" + ParseHelper.FormatDataRow(dataRow) + " ");
logWriter.Flush();
}
}
/// <summary>
///
/// </summary>
/// <param name="pfamId"></param>
public void CalculatePfamLigandComHmmPos(string pfamId)
{
if (ligandComHmmTable == null)
{
InitializeComHmmTable ();
}
string queryString = string.Format("Select PdbID, ChainDomainID, AsymChain, LigandChain, LigandSeqID, SeqID, HmmSeqID From PfamLigands" +
" Where PfamID = '{0}' Order By PdbID, ChainDomainID, LigandChain;", pfamId);
DataTable ligandInteractSeqTable = ProtCidSettings.protcidQuery.Query( queryString);
List<string> entryList = new List<string>();
string pdbId = "";
foreach (DataRow seqRow in ligandInteractSeqTable.Rows)
{
pdbId = seqRow["PdbID"].ToString().TrimEnd();
if (! entryList.Contains (pdbId))
{
entryList.Add(pdbId);
}
}
int numOfComHmmPos = 0;
double jScore = 0;
for (int i = 0; i < entryList.Count; i++)
{
Dictionary<int, List<string>> chainDomainLigandsDictI = GetChainDomainLigandList(entryList[i], ligandInteractSeqTable);
for (int j = i + 1; j < entryList.Count; j++)
{
Dictionary<int, List<string>> chainDomainLigandsDictJ = GetChainDomainLigandList(entryList[j], ligandInteractSeqTable);
foreach (int chainDomainIdI in chainDomainLigandsDictI.Keys)
{
List<string> ligandChainListI = chainDomainLigandsDictI[chainDomainIdI];
foreach (int chainDomainIdJ in chainDomainLigandsDictJ.Keys)
{
List<string> ligandChainListJ = chainDomainLigandsDictJ[chainDomainIdJ];
foreach (string ligandChainI in ligandChainListI)
{
List<int> hmmPosListI = GetLigandInteractingPfamHmmList(entryList[i], chainDomainIdI, ligandChainI, ligandInteractSeqTable);
foreach (string ligandChainJ in ligandChainListJ)
{
List<int> hmmPosListJ = GetLigandInteractingPfamHmmList(entryList[j], chainDomainIdJ, ligandChainJ, ligandInteractSeqTable);
jScore = CalculateJaccardScore(hmmPosListI, hmmPosListJ, out numOfComHmmPos);
if (numOfComHmmPos > 0)
{
DataRow comHmmRow = ligandComHmmTable.NewRow();
comHmmRow["PfamId"] = pfamId;
comHmmRow["PdbID1"] = entryList[i];
comHmmRow["ChainDomainID1"] = chainDomainIdI;
comHmmRow["LigandChain1"] = ligandChainI;
comHmmRow["PdbID2"] = entryList[j];
comHmmRow["ChainDomainID2"] = chainDomainIdJ;
comHmmRow["LigandChain2"] = ligandChainJ;
comHmmRow["NumOfHmmSites1"] = hmmPosListI.Count;
comHmmRow["NumOfHmmSites2"] = hmmPosListJ.Count;
comHmmRow["NumOfComHmmSites"] = numOfComHmmPos;
comHmmRow["Jscore"] = jScore;
ligandComHmmTable.Rows.Add(comHmmRow);
} // if there is shared interacting Pfam HMM positions
}// end of ligandChainJ
} // end of ligandChainI
} // end of chainDomainJ
} // end of chainDomainI
} // end of entrylist J
dbInsert.BatchInsertDataIntoDBtables(ProtCidSettings.protcidDbConnection, ligandComHmmTable);
ligandComHmmTable.Clear();
} // end of entryList I
}
/// <summary>
/// update pdb_entry table
/// </summary>
public void CreatePdbEntryTable()
{
string queryString = "Select Upper(PdbEntry.PdbID) As pdb_id, title, descript as description, method as expdta, keywords, keywords_text, " +
" cast(depositfiledate as varchar(10)) as pdb_ori_date, resolution From PdbEntry, BamEntryKeywords " +
" where PdbEntry.PdbID = BamEntryKeywords.PdbID;";
DataTable pdbEntryTable = dbQuery.Query(ProtCidSettings.pdbfamDbConnection, queryString);
pdbEntryTable.TableName = "pdb_entry";
dbInsert.BatchInsertDataIntoDBtables(bamDbConnect, pdbEntryTable);
}