/// <summary>
///
/// </summary>
/// <param name="auto_pfama_reg_full"></param>
/// <param name="pfamAlignInfoHash"></param>
/// <returns></returns>
private PfamAlignInfo GetPfamAlignInfo(int auto_pfama_reg_full, ref Dictionary <int, PfamAlignInfo> pfamAlignInfoHash)
{
if (pfamAlignInfoHash.ContainsKey(auto_pfama_reg_full))
{
return(pfamAlignInfoHash[auto_pfama_reg_full]);
}
string queryString = string.Format("Select * From PfamA_Reg_Full_Significant Where Auto_PfamA_Reg_full = {0};",
auto_pfama_reg_full);
DataTable alignInfoTable = dbQuery.Query(mysqlConnect, queryString);
PfamAlignInfo pfamAlignInfo = new PfamAlignInfo();
if (alignInfoTable.Rows.Count > 0)
{
pfamAlignInfo.align_start = alignInfoTable.Rows[0]["ali_start"].ToString();
pfamAlignInfo.align_end = alignInfoTable.Rows[0]["ali_end"].ToString();
pfamAlignInfo.model_start = alignInfoTable.Rows[0]["model_start"].ToString();
pfamAlignInfo.model_end = alignInfoTable.Rows[0]["model_end"].ToString();
pfamAlignInfo.domain_bits_score = alignInfoTable.Rows[0]["domain_bits_score"].ToString();
pfamAlignInfo.domain_evalue_score = alignInfoTable.Rows[0]["domain_evalue_score"].ToString();
pfamAlignInfo.sequence_bits_score = alignInfoTable.Rows[0]["sequence_bits_score"].ToString();
pfamAlignInfo.sequence_evalue_score = alignInfoTable.Rows[0]["sequence_evalue_score"].ToString();
}
pfamAlignInfoHash.Add(auto_pfama_reg_full, pfamAlignInfo);
return(pfamAlignInfo);
}
/// <summary>
///
/// </summary>
/// <param name="resultXml"></param>
public PfamAlignInfo[] ParsePfamResultXmlFile(string resultXml)
{
XmlDocument xmlDoc = new XmlDocument();
xmlDoc.Load(resultXml);
// Create an XmlNamespaceManager for resolving namespaces.
XmlNamespaceManager nsManager = new XmlNamespaceManager(xmlDoc.NameTable);
nsManager.AddNamespace("pfam", "http://pfam.sanger.ac.uk/");
XmlNode resultsNode = xmlDoc.DocumentElement.FirstChild;
XmlNode proteinNode = resultsNode.FirstChild;
if (proteinNode == null)
{
throw new Exception("No Pfam alignments found.");
}
XmlNode dbNode = proteinNode.FirstChild;
XmlNode matchesNode = dbNode.FirstChild;
XmlNodeList matchsNodeList = matchesNode.ChildNodes;
ArrayList alignInfoList = new ArrayList();
string pfamAcc = "";
string pfamId = "";
string type = "";
string pfamClass = "";
foreach (XmlNode matchNode in matchesNode.ChildNodes)
{
pfamAcc = matchNode.Attributes["accession"].InnerText;
pfamId = matchNode.Attributes["id"].InnerText;
type = matchNode.Attributes["type"].InnerText;
pfamClass = matchNode.Attributes["class"].InnerText;
foreach (XmlNode locationNode in matchNode.ChildNodes)
{
PfamAlignInfo alignInfo = new PfamAlignInfo();
alignInfo.pfamAcc = pfamAcc;
alignInfo.pfamId = pfamId;
alignInfo.type = type;
alignInfo.pfamClass = pfamClass;
alignInfo.startPos = Convert.ToInt32(locationNode.Attributes["start"].InnerText);
alignInfo.endPos = Convert.ToInt32(locationNode.Attributes["end"].InnerText);
alignInfo.hmmStartPos = Convert.ToInt32(locationNode.Attributes["hmm_start"].InnerText);
alignInfo.hmmEndPos = Convert.ToInt32(locationNode.Attributes["hmm_end"].InnerText);
alignInfo.bitScore = Convert.ToDouble(locationNode.Attributes ["bitscore"].InnerText);
alignInfo.evalue = Convert.ToDouble(locationNode.Attributes["evalue"].InnerText);
alignInfo.mode = locationNode.Attributes["mode"].InnerText;
alignInfoList.Add(alignInfo);
}
}
PfamAlignInfo[] alignInfos = new PfamAlignInfo [alignInfoList.Count];
alignInfoList.CopyTo(alignInfos);
return(alignInfos);
}
public void PrintPfamHmmAlignmentInfo()
{
if (!mysqlConnect.IsConnected())
{
string connectString = "Driver={MySQL ODBC 5.1 Driver}; Server=localhost;Database=pfam;UID=root;PWD=;";
mysqlConnect.ConnectString = connectString;
mysqlConnect.ConnectToDatabase();
}
StreamWriter dataWriter = new StreamWriter(Path.Combine(pfamDataDir, "PdbSeqHmmAlignInfo.txt"));
string dataLine = "";
string queryString = "Select * From pdb_pfamA_reg;";
DataTable pdbChainDomainTable = dbQuery.Query(mysqlConnect, queryString);
int auto_pfamseq = 0;
int auto_pfamA_reg_full = 0;
int auto_pfama = 0;
Dictionary <int, string[]> pfamDefHash = new Dictionary <int, string[]>();
Dictionary <int, string[]> pfamSeqInfoHash = new Dictionary <int, string[]>();
Dictionary <int, PfamAlignInfo> pfamAlignInfoHash = new Dictionary <int, PfamAlignInfo>();
foreach (DataRow chainRow in pdbChainDomainTable.Rows)
{
auto_pfamseq = Convert.ToInt32(chainRow["auto_pfamseq"].ToString());
auto_pfama = Convert.ToInt32(chainRow["auto_pfama"].ToString());
auto_pfamA_reg_full = Convert.ToInt32(chainRow["Auto_Pfama_reg_full"].ToString());
string[] pfamInfo = GetPfamAFamilyInfo(auto_pfama, ref pfamDefHash);
string[] pfamSeqInfo = GetPfamSequenceInfo(auto_pfamseq, ref pfamSeqInfoHash);
PfamAlignInfo pfamAlignInfo = GetPfamAlignInfo(auto_pfamA_reg_full, ref pfamAlignInfoHash);
dataLine = chainRow["pdb_id"].ToString() + "\t" + chainRow["Chain"].ToString().TrimEnd() + "\t" +
pfamInfo[0] + "\t" + pfamInfo[1] + "\t" + pfamInfo[2] + "\t" +
pfamSeqInfo[0] + "\t" + pfamSeqInfo[1] + "\t" + chainRow["pdb_res_start"].ToString() + "\t" +
chainRow["pdb_res_end"].ToString() + "\t" + chainRow["seq_start"].ToString() + "\t" +
chainRow["seq_end"].ToString() + "\t" + pfamAlignInfo.align_start + "\t" + pfamAlignInfo.align_end + "\t" +
pfamAlignInfo.model_start + "\t" + pfamAlignInfo.model_end + "\t" + pfamAlignInfo.domain_bits_score + "\t" +
pfamAlignInfo.domain_evalue_score + "\t" + pfamAlignInfo.sequence_bits_score + "\t" +
pfamAlignInfo.sequence_evalue_score;
dataWriter.WriteLine(dataLine);
}
dataWriter.Close();
mysqlConnect.DisconnectFromDatabase();
}