/// <summary>
///
/// </summary>
/// <param name="alignInfo"></param>
/// <returns></returns>
private string[] GetChainSequences(AlignSeqInfo alignInfo, DataTable seqTable)
{
DataRow[] seqRows = seqTable.Select(string.Format("PdbID = '{0}' AND AsymID = '{1}'",
alignInfo.pdbId, alignInfo.asymChainId));
if (seqRows.Length == 0)
{
seqRows = seqTable.Select(string.Format("PdbID = '{0}' AND AuthorChain = '{1}'",
alignInfo.pdbId, alignInfo.chainId));
if (seqRows.Length == 0)
{
return(null);
}
}
string[] chainSequences = new string[2];
chainSequences[0] = seqRows[0]["SequenceInCoord"].ToString();
chainSequences[1] = seqRows[0]["Sequence"].ToString();
return(chainSequences);
}
/// <summary>
/// Add residues with no-coordinates or no-Calpha to the alignment
/// </summary>
/// <param name="alignInfo1"></param>
/// <param name="alignInfo2"></param>
public void AddDisorderResiduesToAlignment(ref AlignSeqInfo alignInfo1, ref AlignSeqInfo alignInfo2)
{
List <string> pdbList = new List <string> ();
pdbList.Add(alignInfo1.pdbId);
if (!pdbList.Contains(alignInfo2.pdbId))
{
pdbList.Add(alignInfo2.pdbId);
}
DataTable seqTable = GetSequenceTable(pdbList, asuSeqInfoTable);
string[] chainSequences1 = GetChainSequences(alignInfo1, seqTable);
string[] chainSequences2 = GetChainSequences(alignInfo2, seqTable);
// no disorder residues in the middle of the chain
if (!IsSequenceWithDisorderResidues(chainSequences1[0]) && !IsSequenceWithDisorderResidues(chainSequences2[0]))
{
return;
}
try
{
if (HasMissingResidues(alignInfo1.alignSequence, alignInfo1.alignStart, alignInfo1.alignEnd))
{
int[] alignXmlSeqIndexes1 = GetXmlSeqIndexes(ref alignInfo1, chainSequences1[0]);
FillMissingResidues(ref alignInfo1, alignXmlSeqIndexes1, chainSequences1[1], ref alignInfo2);
}
}
catch (Exception ex)
{
throw ex;
}
try
{
if (HasMissingResidues(alignInfo2.alignSequence, alignInfo2.alignStart, alignInfo2.alignEnd))
{
int[] alignXmlSeqIndexes2 = GetXmlSeqIndexes(ref alignInfo2, chainSequences2[0]);
FillMissingResidues(ref alignInfo2, alignXmlSeqIndexes2, chainSequences2[1], ref alignInfo1);
}
}
catch (Exception ex)
{
throw ex;
}
}
/// <summary>
/// residue numbers to xml numbers
/// </summary>
/// <param name="alignInfo"></param>
/// <param name="seqTable"></param>
/// <returns></returns>
private int[] GetXmlSeqIndexes(ref AlignSeqInfo alignInfo, string coordSequence)
{
string nonGapAlignString = GetNonGapSequenceString(alignInfo.alignSequence);
int[] xmlSeqIndexes = GetXmlIndexes(nonGapAlignString, coordSequence);
/* if (xmlSeqNumbers.Length == 0)
* {
* since I used XML sequential numbers, no Blast needed.
* xmlSeqNumbers = MatchSequencesByBlast(nonGapAlignString, coordSequence);
* }*/
if (xmlSeqIndexes.Length == 0)
{
return(null);
}
alignInfo.alignStart = xmlSeqIndexes[0] + 1; // add 1 on January 4, 2017
alignInfo.alignEnd = xmlSeqIndexes[xmlSeqIndexes.Length - 1] + 1;
return(xmlSeqIndexes);
}
/// <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>
/// Fill out missing residues in aligned sequence
/// </summary>
/// <param name="alignInfo1">alignment info for first aligned chain</param>
/// <param name="alignXmlSeq1">xml sequential numbers for first aligned chain</param>
/// <param name="seqString">xml sequence for the first aligned chain</param>
/// <param name="alignInfo2">alignment info for the second chain</param>
private void FillMissingResidues(ref AlignSeqInfo alignInfo1, int[] xmlSeqIndexes, string seqString, ref AlignSeqInfo alignInfo2)
{
Dictionary <int, int> xmlSeqAlignIdxHash = new Dictionary <int, int> ();
int startAlignIdx = 0;
int alignIdx = -1;
Array.Sort(xmlSeqIndexes);
int seqIdx = 0;
for (int i = 0; i < xmlSeqIndexes.Length; i++)
{
alignIdx = GetAlignIndex(alignInfo1.alignSequence, i, startAlignIdx, ref seqIdx);
if (alignIdx < 0)
{
throw new Exception("Get aligned index error for " + xmlSeqIndexes[i].ToString());
}
xmlSeqAlignIdxHash.Add(xmlSeqIndexes[i], alignIdx);
startAlignIdx = alignIdx;
}
int endAlignIdx = -1;
int xmlSeqDif = 0;
int alignSeqDif = 0;
try
{
for (int i = 0; i < xmlSeqIndexes.Length - 1; i++)
{
if (xmlSeqIndexes[i + 1] > xmlSeqIndexes[i] + 1)
{
xmlSeqDif = xmlSeqIndexes[i + 1] - xmlSeqIndexes[i];
startAlignIdx = (int)xmlSeqAlignIdxHash[xmlSeqIndexes[i]];
endAlignIdx = (int)xmlSeqAlignIdxHash[xmlSeqIndexes[i + 1]];
alignSeqDif = endAlignIdx - startAlignIdx;
if (xmlSeqDif > alignSeqDif) // need inserted
{
// propogate the difference for following aligned residues
int dif = xmlSeqDif - alignSeqDif;
for (int j = i + 1; j < xmlSeqIndexes.Length; j++)
{
xmlSeqAlignIdxHash[xmlSeqIndexes[j]] = (int)xmlSeqAlignIdxHash[xmlSeqIndexes[j]] + dif;
}
string insertTemp = "";
int k = 0;
while (k < dif)
{
insertTemp += "-";
k++;
}
// place holder
alignInfo1.alignSequence = alignInfo1.alignSequence.Insert(startAlignIdx + 1, insertTemp);
alignInfo2.alignSequence = alignInfo2.alignSequence.Insert(startAlignIdx + 1, insertTemp);
}
// remove gaps first
// fill out by real residue names from asymunit
string missingResidueString = seqString.Substring(xmlSeqIndexes[i] + 1, xmlSeqDif - 1);
alignInfo1.alignSequence = alignInfo1.alignSequence.Remove(startAlignIdx + 1, missingResidueString.Length);
alignInfo1.alignSequence = alignInfo1.alignSequence.Insert(startAlignIdx + 1, missingResidueString);
}
}
}
catch (Exception ex)
{
throw ex;
}
}
