/// <summary>
/// Read input data from database
/// </summary>
/// <param name="smp">
/// <returns></returns>
List<CompoundStructureActivityData> ReadData(
SasMapParms smp)
{
MetaColumn activityMc = smp.EndpointMc;
QueryColumn keyCriteriaQc = smp.KeyCriteriaQc;
AssertMx.IsNotNull(activityMc, "mc");
AssertMx.IsNotNull(keyCriteriaQc, "keyCriteriaQc");
MetaTable mt, mt2;
MetaColumn mc2 = null;
Query q = new Query();
mt = activityMc.MetaTable;
QueryTable qt = new QueryTable(mt);
if (mt.SummarizedExists && !mt.UseSummarizedData)
{ // retrieve summarized data if exists
mt2 = MetaTableCollection.Get(mt.Name + MetaTable.SummarySuffix);
if (mt2 != null)
{
mc2 = mt2.GetMetaColumnByName(activityMc.Name);
if (mc2 == null) mc2 = mt2.GetMetaColumnByLabel(activityMc.Label);
}
if (mc2 != null) // same column available in summarized?
{
mt = mt2;
activityMc = mc2;
}
}
SMP.KeyCriteriaQc.CopyCriteriaToQueryKeyCritera(q);
q.KeyCriteriaDisplay = SMP.KeyCriteriaQc.CriteriaDisplay;
qt.SelectKeyOnly();
QueryColumn qc = qt.GetQueryColumnByName(activityMc.Name);
qc.Selected = true;
q.AddQueryTable(qt);
QueryEngine qe = new QueryEngine();
List<string> keyList = qe.ExecuteQuery(q); // note that keylist may be empty if single-step query
HashSet<string> keySet = new HashSet<string>(StringComparer.OrdinalIgnoreCase);
List<CompoundStructureActivityData> data = new List<CompoundStructureActivityData>();
int rowCount = 0;
while (true)
{
object[] vo = qe.NextRow();
if (vo == null) break;
CompoundStructureActivityData cd = new CompoundStructureActivityData();
string cid = (string)vo[0];
cd.Cid = cid;
keySet.Add(cid); // accumulate keys
object val = vo[2];
if (NullValue.IsNull(val)) continue;
if (val is double)
cd.Activity = (double)val;
else if (val is Int32)
cd.Activity = (Int32)val;
else if (val is NumberMx)
{
NumberMx nex = val as NumberMx;
cd.Activity = nex.Value;
}
else if (val is QualifiedNumber)
{
QualifiedNumber qn = val as QualifiedNumber;
cd.Activity = qn.NumberValue;
//if (qn.Qualifier != null && qn.Qualifier != "" && qn.Qualifier != "=")
// continue; // (don't want to do this since may filter out good data (e.g. IC50 <0.0001))
}
else continue;
if (cd.Activity == NullValue.NullNumber) continue;
data.Add(cd);
rowCount++;
}
// Retrieve structures
keyList = new List<string>(keySet);
Dictionary<string, MoleculeMx> csDict = MoleculeUtil.SelectMoleculesForCidList(keyList, qt.MetaTable); // get the structures in a single step
// Add structures and build/store fingerprints to data
DebugLog.Message("========== Fingerprints ============");
foreach (CompoundStructureActivityData cd in data)
{
if (!csDict.ContainsKey(cd.Cid) || csDict[cd.Cid] == null) continue;
if (cd.Cid == "111" || cd.Cid == "222") csDict = csDict; // debug
MoleculeMx cs = csDict[cd.Cid];
cd.Structure = cs;
FingerprintType fpType = FingerprintType.Circular;
int fpSubtype = -1;
if (SMP.SimilarityType == SimilaritySearchType.ECFP4) // some issue with ECFP4?
{
fpType = FingerprintType.Circular;
fpSubtype = CircularFingerprintType.ECFP4;
}
else if (SMP.SimilarityType == SimilaritySearchType.Normal)
{
fpType = FingerprintType.MACCS;
}
cd.BitsetFingerprint = cs.BuildBitSetFingerprint(fpType, fpSubtype);
if (cd.BitsetFingerprint == null) continue; // couldn't build fingerprint (e.g. no structure)
if (Debug) DebugLog.Message(cd.Cid + ": " + Lex.Join(CdkMolUtil.GetBitSet(cd.BitsetFingerprint), ", "));
}
return data;
}