/// <summary>
/// Read the list of compounds for a library
/// </summary>
/// <param name="libId"></param>
/// <returns></returns>
public static CidList ReadLibrary(
int libId)
{
CidList l = new CidList();
DbCommandMx dao = new DbCommandMx();
string sql = @"
SELECT l.library_name, l.library_desc_text, s.corp_nbr
FROM corp_owner.corp_substance s,
corp_owner.corp_library_substance ls,
corp_owner.corp_library l
WHERE
l.lib_id = <libId> and
s.cpd_id = ls.cpd_id and
l.lib_id = ls.lib_id" ;
sql = Lex.Replace(sql, "<libId>", libId.ToString());
dao.Prepare(sql);
dao.ExecuteReader();
MetaTable rootMt = MetaTableCollection.Get(MetaTable.PrimaryRootTable);
while (dao.Read())
{
if (Lex.IsNullOrEmpty(l.UserObject.Name))
{
string name = dao.GetString(0);
if (Lex.IsNullOrEmpty(name) || Lex.IsInteger(name))
{
name = dao.GetString(1); // use desc if no name or just a number
}
if (Lex.IsNullOrEmpty(name))
{
name = "Library " + libId;
}
l.UserObject.Name = name;
}
int intCorpId = dao.GetInt(2);
string corpId = CompoundId.Normalize(intCorpId, rootMt);
l.Add(corpId);
}
dao.CloseReader();
dao.Dispose();
return(l);
}
/// <summary>
/// Get list from control removing blank lines & normalizing the ids
/// </summary>
/// <param name="CidList"></param>
void GetNormalizedListFromControl(CidList CidList)
{
CidList.Clear(); // clear contents, keep name
string[] cna = CidListCtl.Lines;
bool allowDuplicates = !RemoveDuplicates.Checked;
for (int i1 = 0; i1 < cna.Length; i1++)
{
string extCn = cna[i1];
if (extCn.Trim() == "" || extCn.StartsWith("("))
{
continue;
}
string intCid = CompoundId.Normalize(extCn, RootTable);
if (String.IsNullOrEmpty(intCid))
{
intCid = extCn;
}
CidList.Add(intCid, allowDuplicates);
}
return;
}
/// <summary>
/// Search button clicked, process input
/// </summary>
/// <returns></returns>
DialogResult ProcessInput()
{
CidList cidList = null;
StreamReader structureFileReader = null;
string qid; // query identifier, compoundId, file name or sdFile key value
QueryManager qm;
DataTableMx dt;
DataColumn dc;
DataRowMx dr;
//object[] dr; // if using Qe
DialogResult dlgRslt = DialogResult.OK;
Query q = null;
QueryTable qt = null;
QueryColumn simScoreQc, structQc; // query column containing latest query settings
MetaTable mt = null;
MetaColumn keyMc = null, structMc = null, dbSetMc = null, simScoreMc = null, mc;
MetaColumnType storageType;
string txt, tok;
if (DatabasesToSearch.Text == "")
{
MessageBoxMx.ShowError("Databases to search must be defined.");
DatabasesToSearch.Focus();
return(DialogResult.Cancel);
}
// Get list of databases
string[] dba = DatabasesToSearch.Text.Split(',');
List <MetaTable> tables = new List <MetaTable>();
foreach (string dbLabel0 in dba)
{
string dbLabel = dbLabel0.Trim();
RootTable dbInfo = RootTable.GetFromTableLabel(dbLabel);
if (dbInfo == null)
{
MessageBoxMx.ShowError("Can't find database " + DatabasesToSearch.Text);
DatabasesToSearch.Focus();
return(DialogResult.Cancel);
}
mt = MetaTableCollection.Get(dbInfo.MetaTableName);
if (mt == null)
{
MessageBoxMx.ShowError("Unable to locate parent structure table for database: " + DatabasesToSearch.Text);
DatabasesToSearch.Focus();
return(DialogResult.Cancel);
}
if (dbSetMc == null)
{
dbSetMc = mt.DatabaseListMetaColumn;
}
tables.Add(mt);
}
if (dbSetMc == null)
{
throw new Exception("\"Databases\" metacolumn not found for any of the databases to search");
}
// Validate other form values
RetrieveStructures = RetrieveMatchingStructures.Checked;
bool fromList = FromList.Checked;
int listCidsRead = 0;
int inputQueryCount = -1;
if (fromList) // using list, validate list name
{
if (SavedListUo == null)
{
MessageBoxMx.ShowError("Compound list must be defined.");
ListName.Focus();
return(DialogResult.Cancel);
}
cidList = CidListCommand.Read(SavedListUo);
if (cidList == null)
{
MessageBoxMx.ShowError("Error reading list.");
ListName.Focus();
return(DialogResult.Cancel);
}
inputQueryCount = cidList.Count;
}
else // Using SdFile, validate SdFile name
{
StructureFile = FileName.Text;
if (StructureFile == "")
{
MessageBoxMx.ShowError("File must be defined.");
FileName.Focus();
return(DialogResult.Cancel);
}
try { structureFileReader = new StreamReader(StructureFile); structureFileReader.Close(); }
catch (Exception ex)
{
MessageBoxMx.ShowError("Can't read file: " + Lex.Dq(StructureFile));
FileName.Focus();
return(DialogResult.Cancel);
}
keyField = KeyField.Text; // get key, blank to use name in 1st line
inputQueryCount = -1; // don't know how many queries unless we read the file (todo?)
}
tok = ResultsName.Text.Trim(); // name to store results under
if (tok == "")
{
MessageBoxMx.ShowError("A name for the results must be provided.");
ResultsName.Focus();
return(DialogResult.Cancel);
}
if (SubStruct.Checked)
{
Psc.SearchType = StructureSearchType.Substructure;
}
else if (Full.Checked)
{
Psc.SearchType = StructureSearchType.FullStructure;
}
else if (Similarity.Checked)
{
Psc.SearchType = StructureSearchType.MolSim;
}
else
{
throw new Exception("Unrecognized search type");
}
// Write initial log entries
SearchCount++;
string logFileName = ClientDirs.DefaultMobiusUserDocumentsFolder + @"\Multistructure Search " + SearchCount + ".txt";
if (!UIMisc.CanWriteFileToDefaultDir(logFileName))
{
return(DialogResult.Cancel);
}
LogStream = new StreamWriter(logFileName);
if (ResultsUo == null)
{
ResultsUo = new UserObject(UserObjectType.Annotation);
}
ResultsUo.Name = tok;
UserObjectTree.GetValidUserObjectTypeFolder(ResultsUo);
DateTime startTime = DateTime.Now;
WriteToLog("Multiple " + Psc.SearchType + " Search");
WriteToLog("Databases: " + DatabasesToSearch.Text);
WriteToLog("Date: " + startTime);
if (fromList)
{
WriteToLog("Input List: " + SavedListUo.Name);
}
else
{
WriteToLog("Input Structure File: " + StructureFile);
}
WriteToLog("Output List: " + ResultsUo.Name);
WriteToLog("Log File: " + logFileName);
WriteToLog("");
WriteToLog("Query, Match, Score");
int queryCount = 0;
int matchAtLeastOneCount = 0;
MoleculeMx queryStructure = null; // current structure being searched
CidList matchList = new CidList();
List <MatchData> matchData = new List <MatchData>();
if (FromFile.Checked) // open SdFile as required
{
structureFileReader = new StreamReader(StructureFile);
}
// Search of structures one at a time
while (true)
{
if (fromList) // get next structure from list
{
if (listCidsRead >= cidList.Count)
{
break;
}
qid = cidList[listCidsRead].Cid;
listCidsRead++;
if (qid.Trim() == "")
{
continue;
}
if (qid.ToLower().IndexOf(".mol") > 0 || qid.ToLower().IndexOf(".skc") > 0)
{ // file reference
if (!File.Exists(qid))
{
continue;
}
if (qid.ToLower().IndexOf(".mol") > 0)
{
queryStructure = MoleculeMx.ReadMolfile(qid);
}
else
{
queryStructure = MoleculeMx.ReadSketchFile(qid);
}
}
else
{
queryStructure = MoleculeUtil.SelectMoleculeForCid(qid);
}
if (queryStructure == null || queryStructure.AtomCount == 0)
{
continue; // oops
}
}
else // get next structure from input file
{
qid = null;
if (StructureFile.ToLower().EndsWith(".sdf"))
{
List <SdFileField> fList = SdFileDao.Read(structureFileReader);
if (fList == null) // end of sdFile
{
structureFileReader.Close();
break;
}
if (fList.Count == 0)
{
continue;
}
queryStructure = new MoleculeMx(MoleculeFormat.Molfile, fList[0].Data);
if (queryStructure == null || queryStructure.AtomCount == 0)
{
continue;
}
if (keyField != "") // key field specified?
{
qid = SdFileDao.GetDataField(fList, keyField);
}
else // get name from 1st line of molfile
{
string molFile = fList[0].Data;
int i1 = molFile.IndexOf("\n");
if (i1 == 0)
{
qid = "";
}
else
{
qid = molFile.Substring(0, i1).Trim();
}
}
if (string.IsNullOrEmpty(qid))
{
qid = SdFileDao.GetDataField(fList, "compound_id");
}
}
else // assume smiles file
{
string smiles = structureFileReader.ReadLine();
if (smiles == null) // end of sdFile
{
structureFileReader.Close();
break;
}
smiles = smiles.Trim();
if (smiles.Length == 0)
{
continue;
}
int i1 = smiles.IndexOf(","); // get any preceeding queryId
if (i1 < 0)
{
i1 = smiles.IndexOf("\t");
}
if (i1 >= 0)
{
qid = smiles.Substring(0, i1).Trim();
smiles = smiles.Substring(i1 + 1).Trim();
}
queryStructure = new MoleculeMx(MoleculeFormat.Smiles, smiles);
if (queryStructure == null || queryStructure.AtomCount == 0)
{
continue;
}
}
if (qid == null || qid.Trim() == "")
{
qid = (queryCount + 1).ToString(); // be sure we have a query id
}
}
queryCount++; // count the query
if (queryStructure == null || queryStructure.AtomCount == 0)
{
WriteToLog("Error converting specific structure " + queryCount.ToString() + ", " + qid);
continue;
}
queryStructure.RemoveStructureCaption(); // remove any Mobius-added caption
Psc.Molecule = queryStructure;
string msg =
"Searching Structure: " + queryCount.ToString();
if (inputQueryCount > 0)
{
msg += " of " + inputQueryCount.ToString();
}
msg += "\n" +
"Structures with one or more matches: " + matchAtLeastOneCount.ToString() + "\n" +
"Total Matches: " + matchList.Count.ToString();
Progress.Show(msg);
// Do the search over the list of databases
for (int ti = 0; ti < tables.Count; ti++)
{
mt = tables[ti];
q = new Query(); // build basic query
//q.SingleStepExecution = true; // do in single step (doesn't currently return sim score)
q.ShowStereoComments = false;
qt = new QueryTable(mt);
q.AddQueryTable(qt);
qt.SelectKeyOnly(); // start selecting desired cols
keyMc = mt.KeyMetaColumn;
structMc = mt.FirstStructureMetaColumn;
structQc = qt.GetQueryColumnByName(structMc.Name);
structQc.Selected = RetrieveStructures;
dbSetMc = mt.DatabaseListMetaColumn;
if (dbSetMc == null)
{
throw new Exception("\"Databases\" metacolumn not found for table: " + mt.Label);
}
QueryColumn dbSetQc = qt.GetQueryColumnByName(dbSetMc.Name);
dbSetQc.Selected = true; // select the database name
RootTable root = RootTable.GetFromTableName(mt.Name);
txt = " in (" + root.Label + ")";
dbSetQc.Criteria = dbSetMc.Name + txt;
dbSetQc.CriteriaDisplay = txt;
simScoreMc = mt.SimilarityScoreMetaColumn;
simScoreQc = null;
if (simScoreMc != null) // get sim score if it exists
{
simScoreQc = qt.GetQueryColumnByName(simScoreMc.Name);
simScoreQc.Selected = true; // return sim score
}
Psc.QueryColumn = structQc;
ParsedStructureCriteria psc2 = AdjustSearchForSmallWorldAsNeeded(Psc);
psc2.ConvertToQueryColumnCriteria(structQc); // format the QC for the structure search
DateTime t0 = DateTime.Now;
//QueryEngine qe = new QueryEngine();
//qe.NextRowsMin = 1000; // minimum number of rows to prefetch
//qe.NextRowsMax = -1; // maximum number of rows to prefetch
//qe.NextRowsMaxTime = 10000; // max time in milliseconds for next fetch
//qe.ExecuteQuery(q);
qm = new QueryManager();
try { dlgRslt = qm.ExecuteQuery(ref q); }
catch (Exception ex)
{
WriteToLog("Error searching structure: " + ex.Message + ", " + queryCount.ToString() + ", " + qid);
continue;
}
if (dlgRslt != DialogResult.OK)
{
return(dlgRslt);
}
double executeTime = TimeOfDay.Delta(ref t0);
int offset = qm.DataTableManager.KeyValueVoPos + 1;
//int offset = 0; // for QE
int keyPos = offset++;
int strPos = RetrieveStructures ? offset++ : -1;
int dbPos = offset++;
int simPos = offset++;
int fetchCnt = 0;
while (true)
{
dr = qm.NextRow();
//dr = qe.NextRow(); // for Qe
if (dr == null)
{
break;
}
fetchCnt++;
if (fetchCnt == 1)
{
matchAtLeastOneCount++; // number of queries that have at least one match
}
MatchData md = new MatchData();
md.Qno = queryCount;
md.Qid = qid;
if (RetrieveStructures)
{
md.Qstr = "Chime=" + queryStructure.GetChimeString();
}
CompoundId cid = CompoundId.ConvertTo(dr[keyPos]);
md.Mid = cid.Value;
if (RetrieveStructures)
{
MoleculeMx ms = MoleculeMx.ConvertTo(dr[strPos]);
if (!NullValue.IsNull(ms))
{
md.Mstr = "Chime=" + ms.GetChimeString();
}
}
StringMx db = StringMx.ConvertTo(dr[dbPos]);
if (!NullValue.IsNull(db))
{
md.Db = db.Value;
}
if (Psc.SearchType == StructureSearchType.MolSim)
{
NumberMx nex = NumberMx.ConvertTo(dr[simPos]);
if (!NullValue.IsNull(nex))
{
md.Score = nex.Value;
}
}
if (matchList.Contains(cid.Value)) // already have compound id as match for other query?
{
if (Psc.SearchType != StructureSearchType.MolSim)
{
continue; // if similarity search see if more similar
}
CidListElement le = matchList.Get(cid.Value); // reference current score
if (le.Tag > md.Score)
{
continue; // only replace if more similar
}
matchList.Remove(le.Cid); // remove from list
for (int mi = 0; mi < matchData.Count; mi++) // remove from data
{
if (matchData[mi].Mid == md.Mid)
{
matchData.RemoveAt(mi);
break;
}
}
}
matchList.Add(md.Mid);
matchList.Get(md.Mid).Tag = md.Score; // keep score in list
matchData.Add(md); // add to results
txt = md.Qid + ", " + md.Mid + ", " + md.Score.ToString();
WriteToLog(txt);
} // Fetch result loop
double fetchTime = TimeOfDay.Delta(ref t0);
} // DB loop
if (Progress.CancelRequested)
{
Progress.Hide();
MessageBoxMx.ShowError("Search cancelled.");
try { LogStream.Close(); } catch { }
return(DialogResult.Cancel);
}
} // key loop
CidListCommand.WriteCurrentList(matchList); // write the list of numbers
UsageDao.LogEvent("MultipleStructSearch");
txt =
"=== Multiple structure search complete ===\r\n\r\n" +
"Structures Searched: " + queryCount.ToString() + "\r\n";
txt +=
"Structures with one or more matches: " + matchAtLeastOneCount.ToString() + "\r\n" +
"Total Matches: " + matchList.Count.ToString() + "\r\n";
TimeSpan ts = DateTime.Now.Subtract(startTime);
ts = new TimeSpan(ts.Hours, ts.Minutes, ts.Seconds);
txt += "Total Time: " + ts + "\r\n\r\n";
WriteToLog("\r\n" + txt);
try { LogStream.Close(); } catch { }
if (matchList.Count == 0)
{
Progress.Hide();
MessageBoxMx.ShowError("No matches have been found.");
return(DialogResult.Cancel);
}
tok = "Matching compound ids";
if (Psc.SearchType == StructureSearchType.MolSim)
{
tok = "Similar compound ids";
}
txt += tok + " have been written to the current list: " + ResultsUo.Name + "\n" +
"Log file written to: " + logFileName + "\n\n" +
"Do you want to view the match results?";
DialogResult dRslt = MessageBoxMx.Show(txt, "Multiple Structure Search", MessageBoxButtons.YesNoCancel, MessageBoxIcon.Question);
if (dRslt == DialogResult.Cancel)
{
return(DialogResult.Cancel);
}
else if (dRslt == DialogResult.No) // show log
{
SystemUtil.StartProcess(logFileName);
return(DialogResult.Cancel);
}
// Display results
Progress.Show("Formatting results...");
mt = new MetaTable();
mt.Name = "MULTSTRUCTSEARCH_" + SearchCount;
mt.Label = "Multiple Structure Search " + SearchCount;
mt.TableMap = "Mobius.Tools.MultStructSearch"; // plugin id
mt.MetaBrokerType = MetaBrokerType.NoSql;
ColumnSelectionEnum structureColumnSelection = RetrieveStructures ? ColumnSelectionEnum.Selected : ColumnSelectionEnum.Unselected;
keyMc = keyMc.Clone();
keyMc.Name = "MatchingCid";
keyMc.Label = "Matching Compound Id";
mt.AddMetaColumn(keyMc);
structMc = structMc.Clone();
structMc.Name = "MatchingStructure";
structMc.Label = "Matching Structure";
structMc.InitialSelection = structureColumnSelection;
mt.AddMetaColumn(structMc);
dbSetMc = dbSetMc.Clone();
dbSetMc.Name = "Database";
mt.AddMetaColumn(dbSetMc);
//if (DatabasesToSearch.Text.Contains(","))
dbSetMc.InitialSelection = ColumnSelectionEnum.Selected;
mc = mt.AddMetaColumn("Molsimilarity", "Similarity Search Score", MetaColumnType.Number, ColumnSelectionEnum.Unselected, 10);
if (Psc.SearchType == StructureSearchType.MolSim)
{
mc.InitialSelection = ColumnSelectionEnum.Selected;
}
mc = mt.AddMetaColumn("QueryNo", "Query Number", MetaColumnType.Integer);
//mc = mt.AddMetaColumn("QueryMatchNo", "Query Match Number", MetaColumnType.Integer);
mc = mt.AddMetaColumn("QueryId", "Query Id", MetaColumnType.String);
mc = mt.AddMetaColumn("QueryStructure", "Query Structure", MetaColumnType.Structure);
mc.InitialSelection = structureColumnSelection;
q = ToolHelper.InitEmbeddedDataToolQuery(mt);
dt = q.ResultsDataTable as DataTableMx;
for (int mi = 0; mi < matchData.Count; mi++)
{
MatchData md = matchData[mi];
dr = dt.NewRow();
dr[qt.Alias + ".MatchingCid"] = new CompoundId(md.Mid);
if (RetrieveStructures)
{
dr[qt.Alias + ".MatchingStructure"] = new MoleculeMx(MoleculeFormat.Chime, md.Mstr);
}
dr[qt.Alias + ".Database"] = new StringMx(md.Db);
if (Psc.SearchType == StructureSearchType.MolSim)
{
dr[qt.Alias + ".Molsimilarity"] = new NumberMx(md.Score);
}
dr[qt.Alias + ".QueryNo"] = new NumberMx(md.Qno);
dr[qt.Alias + ".QueryId"] = new StringMx(md.Qid);
if (RetrieveStructures)
{
dr[qt.Alias + ".QueryStructure"] = new MoleculeMx(MoleculeFormat.Chime, md.Qstr);
}
dt.Rows.Add(dr);
}
ToolHelper.DisplayData(q, dt, true);
Progress.Hide();
return(DialogResult.OK);
}
/// <summary>
/// Attempt to read existing results file into the query DataTable
/// </summary>
/// <param name="qm"></param>
public void ReadBinaryResultsFile(string fileName)
{
QueryTable qt;
QueryColumn qc;
BinaryReader br = null;
Stopwatch sw = Stopwatch.StartNew();
try
{
bool saveHandlersEnabled = Qm.DataTable.EnableDataChangedEventHandlers(false); // disable for faster load
bool saveUpdateMaxRowsPerKey = UpdateMaxRowsPerKeyEnabled;
UpdateMaxRowsPerKeyEnabled = false; // disable for faster load
int id = Query.UserObject.Id;
if (id <= 0)
{
throw new Exception("Query not saved");
}
if (DataTableMx == null || DataTableMx.Columns.Count == 0)
{
throw new Exception("DataTable not defined");
}
br = BinaryFile.OpenReader(fileName);
string sq = br.ReadString();
Query q0 = Query.Deserialize(sq); // deserialize the saved query
QueryManager qm0 = new QueryManager();
qm0.LinkMember(q0);
ResultsFormat rf0 = new ResultsFormat(qm0, OutputDest.WinForms);
ResultsFormatFactory rff0 = new ResultsFormatFactory(qm0, OutputDest.WinForms);
rff0.Build(); // build format with vo positions
// The cached query cols should match those of the current query: however,
// we'll create a mapping just in case they don't
int voArrayLen0 = br.ReadInt32(); // cached vo array len
int voArrayLen = DataTableMx.Columns.Count - KeyValueVoPos; // current query vo array len
List <int> q0VoMap = new List <int>(); // vo position in cached query data
List <int> qVoMap = new List <int>(); // vo position in current version of query
q0VoMap.Add(0); // first position is the common key value
qVoMap.Add(0);
foreach (QueryTable qt0 in q0.Tables) // scan each table in cached data
{
foreach (QueryColumn qc0 in qt0.QueryColumns) // and each column
{
if (qc0.VoPosition < 0)
{
continue; // skip if not mapped to the vo in cached data
}
int q0VoPos = qc0.VoPosition - KeyValueVoPos; // where it is in cache
int qvoPos = -1; // where it will go
qt = Query.GetTableByName(qt0.MetaTable.Name);
if (qt != null)
{
qc = qt.GetQueryColumnByName(qc0.MetaColumn.Name);
if (qc != null)
{
qvoPos = qc.VoPosition - KeyValueVoPos;
}
}
q0VoMap.Add(q0VoPos); // where it is in saved data
qVoMap.Add(qvoPos); // where it will go (not including attributes & check cols)
}
}
if (q0VoMap.Count != voArrayLen0)
{
throw new Exception("Cached Vo length doesn't match list of selected columns");
}
DataTableMx.Clear(); // clear the rows
CidList cidList = new CidList();
object[] voa = new object[voArrayLen]; // array to fill
while (!BinaryFile.ReaderEof(br)) // process each row
{
for (int mi = 0; mi < q0VoMap.Count; mi++) // each col
{
object o = VoArray.ReadBinaryItem(br);
if (mi == 0 && o != null) // add to key list if key
{
cidList.Add(o.ToString(), false);
}
if (qVoMap[mi] >= 0) // save in new buf if mapped
{
voa[qVoMap[mi]] = o;
}
}
DataRowMx dr = AddDataRow(voa);
}
br.Close();
Qm.DataTable.EnableDataChangedEventHandlers(saveHandlersEnabled);
UpdateMaxRowsPerKeyEnabled = saveUpdateMaxRowsPerKey;
InitializeRowAttributes(false);
ResultsKeys = cidList.ToStringList(); // include keys in DTM as well
double ms = sw.Elapsed.TotalMilliseconds;
return;
}
catch (Exception ex)
{
if (br != null)
{
br.Close();
}
throw new Exception(ex.Message, ex);
}
}
