private Query CreateLuceneQuery(ChemicalStructure queryStructure, SearchType searchType)
{
BooleanQuery query = new BooleanQuery();
switch (searchType)
{
// For exact searches, the search results MUST match the exact key.
case SearchType.Exact:
query.Add(new TermQuery(new Term(ChemicalStructureIndexer.FIELD_EXACTKEY, queryStructure.GetUniqueKey())), Occur.MUST);
break;
// For similarity searches, the search results SHOULD contain the query similarity fingerprint positions.
case SearchType.Similarity:
foreach (int fingerprintPosition in queryStructure.GetSimilarityFingerprintPositions())
{
query.Add(new TermQuery(new Term(ChemicalStructureIndexer.FIELD_FINGERPRINT_POSITION_SIMILARITY, fingerprintPosition.ToString())), Occur.SHOULD);
}
break;
// For substructure searches, the search results MUST contain ALL of the query substructure fingerprint positions.
case SearchType.Substructure:
foreach (int fingerprintPosition in queryStructure.GetSubstructureFingerprintPositions())
{
query.Add(new TermQuery(new Term(ChemicalStructureIndexer.FIELD_FINGERPRINT_POSITION_SUBSTRUCTURE, fingerprintPosition.ToString())), Occur.MUST);
}
break;
}
return(query);
}
/// <summary>
/// Searches for a specified query structure.
/// </summary>
/// <param name="queryStructure">The chemical structure to search for.</param>
/// <param name="searchType">The desired search type (Exact, Similarity, or Substructure).</param>
/// <returns></returns>
public List <ChemicalStructureSearchResult> Search(ChemicalStructure queryStructure, SearchType searchType)
{
var results = new List <ChemicalStructureSearchResult>();
// Form the lucene query.
Query query = CreateLuceneQuery(queryStructure, searchType);
// Execute to obtain lucene hit pointers; we're going to artifically cap this out at 100 hits.
TopDocs hits = this.IndexSearcher.Search(query, 100);
// Loop through and form the results.
foreach (ScoreDoc scoreDoc in hits.ScoreDocs)
{
// Retrieve the lucene document, and form a chemical structure result.
Document doc = this.IndexSearcher.Doc(scoreDoc.Doc);
var result = new ChemicalStructureSearchResult()
{
ChemicalStructure = new ChemicalStructure(doc.Get(ChemicalStructureIndexer.FIELD_NAME), doc.Get(ChemicalStructureIndexer.FIELD_MOLFILE)),
Score = scoreDoc.Score,
};
// One catch: for Substructure searches, we have actually identified the set of chemical structures that MIGHT be
// substructure matches; for Substructure searches, we need to perform an actual substructure determination.
if ((searchType != SearchType.Substructure) || result.ChemicalStructure.HasSubstructure(queryStructure))
{
results.Add(result);
}
}
return(results);
}
/// <summary>
/// Returns whether the chemical structure contains a specified substructure.
/// </summary>
/// <param name="substructureQuery">The specified substructure to search for.</param>
/// <returns>True if this chemical structure contains the specified substructure.</returns>
public bool HasSubstructure(ChemicalStructure substructureQuery)
{
bool hasSubstructure = false;
using (Indigo indigo = new Indigo())
{
// Load inputs.
IndigoObject structure = CreateIndigoStructure(indigo);
IndigoObject substructure = indigo.loadQueryMolecule(substructureQuery.MolfileContents);
// Perform the match.
IndigoObject substructureMatcher = indigo.substructureMatcher(structure);
hasSubstructure = (substructureMatcher.match(substructure) != null);
// Dispose.
structure.Dispose();
substructure.Dispose();
substructureMatcher.Dispose();
}
return(hasSubstructure);
}
/// <summary>
/// Adds a chemical structure to the index.
/// </summary>
/// <param name="chemicalStructure">The chemical structure to add to the index.</param>
public void AddChemicalStructure(ChemicalStructure chemicalStructure)
{
// Create a new Lucene document for the chemical structure.
Document doc = new Document();
// Add stored fields for the chemical structure name and molfile; this will
// allow us to retrieve them later.
doc.Add(new Field(FIELD_NAME, chemicalStructure.Name, Field.Store.YES, Field.Index.NO));
doc.Add(new Field(FIELD_MOLFILE, chemicalStructure.MolfileContents, Field.Store.YES, Field.Index.NO));
// Add searchable fields for exact key and similarity/substructure fingerprints.
doc.Add(new Field(FIELD_EXACTKEY, chemicalStructure.GetUniqueKey(), Field.Store.NO, Field.Index.NOT_ANALYZED));
foreach (int fingerprintPosition in chemicalStructure.GetSimilarityFingerprintPositions())
{
doc.Add(new Field(FIELD_FINGERPRINT_POSITION_SIMILARITY, fingerprintPosition.ToString(), Field.Store.NO, Field.Index.NOT_ANALYZED));
}
foreach (int fingerprintPosition in chemicalStructure.GetSubstructureFingerprintPositions())
{
doc.Add(new Field(FIELD_FINGERPRINT_POSITION_SUBSTRUCTURE, fingerprintPosition.ToString(), Field.Store.NO, Field.Index.NOT_ANALYZED));
}
// Add the document to the index.
this.IndexWriter.AddDocument(doc);
}
