/// <summary>
/// Begin a transaction
/// </summary>
public void BeginTransaction()
{
if (DbCmd == null || DbCmd.MxConn == null || DbCmd.MxConn.DbConn == null)
{
Connect(); // be sure we have a connection
}
DbCmd.BeginTransaction();
return;
}
/// <summary>
/// Copy PRD Annotation Table data to DEV by groups of result ids
/// </summary>
/// <returns></returns>
public static string CopyPrdAnnotationTableDataToDevBy_Rslt_Id()
{
long minRi = 1;
//long minRi = 1310251601; // appro half of rows, allows index on all cols to be built
long maxRi = 4331096190;
long chunkSize = 100000;
DbCommandMx cmd = new DbCommandMx();
cmd.MxConn = DbConnectionMx.GetConnection("DEV857");
cmd.BeginTransaction();
long totalIns = 0;
for (long ri = minRi; ri <= maxRi; ri += chunkSize)
{
string range = (ri + 1).ToString() + " and " + (ri + chunkSize);
string sql =
@"insert /*+ APPEND */ into mbs_owner.mbs_adw_rslt
(select * from mbs_owner.mbs_adw_rslt
where rslt_id between " + range + ")";
cmd.PrepareUsingDefinedConnection(sql);
int insCnt = cmd.ExecuteNonReader();
cmd.Commit();
totalIns += insCnt;
string msg = "Rows copied: " + totalIns + " (" + range + ")";
Mobius.UAL.Progress.Show(msg);
DebugLog.Message(msg);
}
cmd.Dispose();
return("Rows copied: " + totalIns);
}
/// <summary>
/// /// <summary>
/// Copy PRD Annotation Table data to DEV reording and compressing the data by method
/// </summary>
/// <returns></returns>
public static string CopyPrdAnnotationTableDataToDev()
{
string mvidSql = @"
select mthd_vrsn_id, count from
(
select mthd_vrsn_id, count(mthd_vrsn_id) count
FROM MBS_OWNER.mbs_adw_rslt
WHERE mthd_vrsn_id > 0 /* = 708070 */
group by mthd_vrsn_id)
order by mthd_vrsn_id
";
Mobius.UAL.Progress.Show("Getting list of annotation table mtht_vrsn_ids ...");
DbCommandMx c1 = new DbCommandMx();
c1.Prepare(mvidSql);
c1.ExecuteReader();
DbCommandMx cmd = new DbCommandMx();
cmd.MxConn = DbConnectionMx.GetConnection("DEV857");
cmd.BeginTransaction();
long totalIns = 0;
int vmIdCnt = 0;
while (c1.Read())
{
long mvId = c1.GetLong(0);
string sql =
@"insert /*+ APPEND */ into mbs_owner.mbs_adw_rslt
select * from mbs_owner.mbs_adw_rslt
where mthd_vrsn_id = " + mvId + @"
order by ext_cmpnd_id_nbr, rslt_grp_id, rslt_id";
cmd.PrepareUsingDefinedConnection(sql);
int insCnt = cmd.ExecuteNonReader();
cmd.Commit();
totalIns += insCnt;
vmIdCnt++;
string msg = "Mthd_Vrsn_id: " + mvId + ", Vmids: " + vmIdCnt + ", Rows: " + totalIns;
Mobius.UAL.Progress.Show(msg);
DebugLog.Message(msg);
}
c1.Dispose();
cmd.Dispose();
return("Rows copied: " + totalIns);
}
/// <summary>
/// Update Common Assay Attributes table (mbs_owner.cmn_assy_atrbts)
///
/// Command: Update AssayAttributesTable
///
/// This command builds an entry (or two) in the cmn_assy_atrbts table for each assay
/// referenced in the Mobius contents tree that reports an SP or CRC value as determined
/// by the associated metafactory and available in each metatable.
/// If the gene target associated with an assay can be identified then information on that
/// gene is included as well.
///
/// Additional gene information may come from the metadata for a source such as the results
/// warehouse.
///
/// Note that this function must be run under an account that has access to all restricted data so that it
/// can properly see what's available and build the table.
/// </summary>
/// <param name="lex"></param>
/// <returns></returns>
public static string UpdateAssayAttributesTable(string args)
{
MetaTable mt;
MetaColumn mc;
AssayAttributes aa, aa2;
Dictionary <int, int> geneIdCounts = new Dictionary <int, int>(); // genes and counts keyed by entrez gene id
Dictionary <int, double> targetMapXDict; // dendogram X coord keyed by gene id
Dictionary <int, double> targetMapYDict; // dendogram Y coord keyed by gene id
Dictionary <string, int> targetTypeCounts = new Dictionary <string, int>(); // target types and counts
Dictionary <int, string> assayTypeDict;
Dictionary <int, string> assayModeDict;
//UnpivotedAssayResult rr, rr2;
List <string> toks;
string mtName, tableNamePrefix, key, fileName, msg;
bool crcExists, spExists, isSummary;
int tableId, step, ri, resultTypeId, assayRowCount;
Log = new LogFile(ServicesDirs.LogDir + @"\UpdateCommonAssayAttributes.log");
Log.ResetFile();
// Get list of all of the assays in the tree
LogMessage("Accumulating assays...");
HashSet <string> mtNameHash = new HashSet <string>();
foreach (MetaTreeNode mtn0 in MetaTree.Nodes.Values)
{
if (mtn0.Type != MetaTreeNodeType.MetaTable)
{
continue;
}
mtName = mtn0.Target;
if (AssayMetaData.IsAssayMetaTableName(mtName))
{
mtNameHash.Add(mtName);
}
}
bool debug = false; // set to true to debug with limited list of assays from below
if (debug)
{
mtNameHash.Clear();
//assayHash.Add("ASSAY_1");
//assayHash.Add("ASSAY_2");
//assayHash.Add("ASSAY_3");
}
// Get other informatin needed from AssayMetadata
LogMessage("Reading AssayMetadata ASSAY metadata...");
Dictionary <int, AssayDbMetadata> assayMetadataAssayDict = // get assays and associated target/gene information
AssayMetadataDao.GetAssayTargetGeneData();
LogMessage("Getting AssayMetadata result types...");
Dictionary <int, AssayDbResultType> resultTypeDict = AssayMetadataDao.GetResultTypeDict();
LogMessage("Getting assay types and modes...");
AssayMetadataDao.GetAssayTypesAndModes(out assayTypeDict, out assayModeDict);
LogMessage("Getting gene dendogram coordinates...");
try
{
AssayMetadataDao.GetImageCoords(out targetMapXDict, out targetMapYDict);
}
catch (Exception ex) // may fail if problem with data source
{
LogMessage(DebugLog.FormatExceptionMessage(ex, true));
targetMapXDict = new Dictionary <int, double>();
targetMapYDict = new Dictionary <int, double>();
}
// Process each assay
int metatablesFound = 0, metatablesNotFound = 0;
int assaysProcessed = 0;
int assaysWithGenes = 0;
int assaysWithGeneCoords = 0;
int assaysWithTargets = 0;
int assaysWithSpOnly = 0;
int assaysWithCrcOnly = 0;
int assaysWithNoCrcSP = 0;
int assaysWithOtherTypes = 0;
int assaysWithCrcAndSp = 0;
int assaysWithNoKeyTypes = 0;
int assaysWithProcessingErrors = 0;
Dictionary <string, int> CrcAssayCnt = new Dictionary <string, int>()
{
{ "ASSAY_DB1", 0 }, { "ASSAY_DB2", 0 }, { "ASSAY_DB3", 0 }
};
Dictionary <string, int> SpAssayCnt = new Dictionary <string, int>()
{
{ "ASSAY_DB1", 0 }, { "ASSAY_DB2", 0 }, { "ASSAY_DB3", 0 }
};
Dictionary <string, int> OtherAssayCnt = new Dictionary <string, int>()
{
{ "ASSAY_DB1", 0 }, { "ASSAY_DB2", 0 }, { "ASSAY_DB3", 0 }
};
List <AssayAttributes> resultTypeRows = new List <AssayAttributes>();
List <AssayAttributes> geneRows = new List <AssayAttributes>();
List <AssayAttributes> dbRows = new List <AssayAttributes>();
string copyUpdateMsg = "";
foreach (string mtName0 in mtNameHash)
{
AssayMetaData assayMetaData = null; // metadata for assay
bool isAssay = false;
int assayIdNbr = NullValue.NullNumber;
int assayId = NullValue.NullNumber;
//if (assaysProcessed >= 100) break; // debug
mtName = mtName0;
MetaTable.ParseMetaTableName(mtName, out tableNamePrefix, out tableId, out isSummary);
string resultType = "";
string rtId = "";
string assayName, assayDb;
mt = MetaTableCollection.Get(mtName); // get metatable
if (mt == null)
{
metatablesNotFound++;
LogMessage("MetaTable not found: " + mtName);
continue;
}
metatablesFound++;
if (mt.Code == "")
{
continue; // must be single pivoted assay
}
assayDb = "ASSAY_DB"; // customize
assayIdNbr = -1;
if (UalUtil.IClient != null && UalUtil.IClient.Attended)
{
UAL.Progress.Show((assaysProcessed + 1).ToString() + " / " + mtNameHash.Count + " - " + mt.Name + "\r\n" + mt.Label);
}
aa = new AssayAttributes();
aa.AssayDatabase = assayDb;
aa.AssayIdNbr = assayIdNbr; // data-source-specific assay Id
aa.AssayIdTxt = mt.Name; // store ASSAY_1234 type table name
aa.AssayId2 = assayId; // any associated assay id
if (isAssay)
{
aa.AssayName = assayMetaData.Name; // name from AssayMetadata
}
else
{
aa.AssayName = MetaTable.RemoveSuffixesFromName(mt.Label); // name from metatable
}
if (mt.SummarizedExists)
{
aa.SummarizedAvailable = true;
}
else
{
aa.SummarizedAvailable = false;
}
if (isAssay)
{
if (assayTypeDict.ContainsKey(tableId))
{
aa.AssayType = assayTypeDict[tableId];
}
if (assayModeDict.ContainsKey(tableId))
{
aa.AssayMode = assayModeDict[tableId];
}
aa.AssaySource = AssayMetaData.GetAssaySource(tableId);
aa.AssociationSource = "TODO"; // customize
}
aa.AssayStatus = "Active"; // say all active for now
MetaTableStats mts = MetaTableFactory.GetStats(mtName);
if (mts != null)
{
aa.ResultCount = (int)mts.RowCount;
aa.AssayUpdateDate = mts.UpdateDateTime;
}
else
{
aa.ResultCount = 0; // assume no results if no stats
}
if (mt.DescriptionIsAvailable()) // use description from Mobius
{
aa.AssayDesc = "Y";
}
if (String.IsNullOrEmpty(aa.GeneFamily))
{
aa.GeneFamily = "Unknown"; // set these to "Unknown" rather than null
}
if (String.IsNullOrEmpty(aa.AssayType))
{
aa.AssayType = "UNKNOWN"; // upper case UNKNOWN
}
if (String.IsNullOrEmpty(aa.AssayMode))
{
aa.AssayMode = "UNKNOWN"; // upper case UNKNOWN
}
// Step1: Add a row for primary & any secondary results
resultTypeRows.Clear();
MetaColumn firstResultCol = null, firstKeyResultCol = null, firstOtherKeyResultCol = null;
string resultTypeConcType;
HashSet <string> keyResultTypeCodes = new HashSet <string>();
int spCnt = 0, crcCnt = 0, otherCnt = 0;
for (int mci = 0; mci < mt.MetaColumns.Count; mci++) // pick first col with result code (could also check summarization method)
{
mc = mt.MetaColumns[mci];
if (Lex.IsUndefined(mc.ResultCode))
{
continue; // must have code defined
}
if (keyResultTypeCodes.Contains(mc.ResultCode))
{
continue; // and not included so far
}
if (mc.InitialSelection != ColumnSelectionEnum.Selected)
{
continue; // selected only
}
if (firstResultCol == null)
{
firstResultCol = mc;
}
if (!IsKeyResultType(mc, out resultTypeConcType))
{
continue;
}
if (firstKeyResultCol == null)
{
firstKeyResultCol = mc;
}
keyResultTypeCodes.Add(mc.ResultCode);
aa2 = aa.Clone();
if (resultTypeRows.Count == 0)
{
aa2.TopLevelResult = "Y";
}
else
{
aa2.TopLevelResult = "N";
}
aa2.ResultTypeId2 = GetAssayResultTypeId(mc); // AssayMetadata result type id
aa2.ResultTypeIdNbr = GetInternalResultTypeId(mc); // Internal database result type id
aa2.ResultTypeIdTxt = mc.Name; // Mobius column name
if (isAssay && resultTypeDict.ContainsKey(aa2.ResultTypeId2))
{
aa2.ResultName = resultTypeDict[aa2.ResultTypeId2].Name; // use name from AssayMetadata result type dict
}
else
{
aa2.ResultName = mc.Label; // use label from Mobius
}
aa2.ResultTypeUnits = mc.Units; // result units
if (Lex.Eq(resultTypeConcType, "SP"))
{
aa2.ResultTypeConcType = "SP";
spCnt++;
}
else if (Lex.Eq(resultTypeConcType, "CRC"))
{
aa2.ResultTypeConcType = "CRC";
crcCnt++;
}
else
{
aa2.ResultTypeConcType = "";
otherCnt++;
if (firstOtherKeyResultCol == null)
{
firstOtherKeyResultCol = mc;
}
}
aa2.ResultTypeConcUnits = ""; // todo
resultTypeRows.Add(aa2);
}
if (resultTypeRows.Count >= 1)
{
if (crcCnt > 0)
{
CrcAssayCnt[assayDb]++; // count primary type by db
}
else if (spCnt > 0)
{
SpAssayCnt[assayDb]++;
}
else
{
OtherAssayCnt[assayDb]++;
}
if (crcCnt > 0 && spCnt == 0)
{
assaysWithCrcOnly++; // count overall primary/secondary types
}
else if (crcCnt == 0 && spCnt > 0)
{
assaysWithSpOnly++;
}
else if (crcCnt > 0 && spCnt > 0)
{
assaysWithCrcAndSp++;
}
if (crcCnt == 0 && spCnt == 0) // no SP or CRC result types
{
assaysWithNoCrcSP++;
mc = firstKeyResultCol;
LogMessage("Assay with No SP/CRC key results: " + mt.Name + "." + mc.Name + " (" + mc.ResultCode + "), " + mt.Label + "." + mc.Label);
}
else if (otherCnt > 0) // no SP or CRC result types
{
assaysWithOtherTypes++;
mc = firstOtherKeyResultCol;
LogMessage("Non SP/CRC key result: " + mt.Name + "." + mc.Name + " (" + mc.ResultCode + "), " + mt.Label + "." + mc.Label);
}
}
else // no key result type
{
aa2 = aa.Clone();
resultTypeRows.Add(aa2); // include row for step1
OtherAssayCnt[assayDb]++;
assaysWithNoKeyTypes++;
LogMessage("No key result type for metatable: " + mt.Name + ", " + mt.Label);
}
// Build a step2 row for each target/gene
geneRows.Clear();
List <AssayTarget> targets = new List <AssayTarget>();
int geneCount = 0;
if (isAssay)
{
targets = assayMetaData.Targets;
}
if (targets.Count > 0)
{
assaysWithTargets++;
}
foreach (AssayTarget target in targets)
{
aa = new AssayAttributes();
aa.GeneFamily = target.TargetTypeShortName; // count target type occurance
if (Lex.IsUndefined(aa.GeneFamily))
{
aa.GeneFamily = "Unknown";
}
if (!targetTypeCounts.ContainsKey(aa.GeneFamily))
{
targetTypeCounts[aa.GeneFamily] = 0;
}
targetTypeCounts[aa.GeneFamily]++;
if (target.Genes == null || target.Genes.Count == 0) // if no genes add a single target row
{
geneRows.Add(aa);
continue;
}
foreach (AssayGene rg in target.Genes)
{
if (!Lex.IsDefined(rg.GeneSymbol))
{
continue;
}
aa2 = aa.Clone();
geneRows.Add(aa2);
aa2.GeneSymbol = rg.GeneSymbol;
int.TryParse(rg.GeneId, out aa2.GeneId);
if (aa2.GeneId > 0 && targetMapXDict.ContainsKey(aa2.GeneId))
{
aa2.TargetMapX = targetMapXDict[aa2.GeneId];
aa2.TargetMapY = targetMapYDict[aa2.GeneId];
if (geneCount == 0)
{
assaysWithGeneCoords++;
}
}
if (!geneIdCounts.ContainsKey(aa2.GeneId)) // count gene occurance
{
geneIdCounts[aa2.GeneId] = 0;
}
geneIdCounts[aa2.GeneId]++;
if (geneCount == 0)
{
assaysWithGenes++;
}
geneCount++;
}
}
if (geneRows.Count == 0) // if no step 2 rows (i.e. no targets), create a single step2 row
{
aa = new AssayAttributes();
geneRows.Add(aa);
}
// Combine key result types with target/genes
for (int i1 = 0; i1 < resultTypeRows.Count; i1++)
{
AssayAttributes s1aa = resultTypeRows[i1];
for (int i2 = 0; i2 < geneRows.Count; i2++)
{
AssayAttributes s2aa = geneRows[i2];
aa = s1aa.Clone();
aa.GeneId = s2aa.GeneId;
aa.GeneSymbol = s2aa.GeneSymbol;
aa.GeneFamily = s2aa.GeneFamily;
aa.TargetMapX = s2aa.TargetMapX;
aa.TargetMapY = s2aa.TargetMapY;
aa.GeneCount = geneCount;
if (i2 > 0)
{
aa.GeneCount = -geneCount; // negative for other than 1st gene
}
dbRows.Add(aa);
}
}
assaysProcessed++;
}
// Update table
bool updateTable = true; // set to false for debug
if (dbRows.Count <= 0)
{
LogMessage("No rows in new dataset, table not updated");
}
else if (updateTable)
{
LogMessage("Deleting existing data...");
DbCommandMx dao = new DbCommandMx();
string sql = "delete from mbs_owner.cmn_assy_atrbts";
sql = AssayAttributesDao.AdjustAssayAttrsTableName(sql);
dao.Prepare(sql);
dao.BeginTransaction();
int delCnt = dao.ExecuteNonReader();
LogMessage("Inserting new data...");
int t0 = TimeOfDay.Milliseconds();
for (ri = 0; ri < dbRows.Count; ri++)
{
aa = dbRows[ri];
aa.Id = ri + 1;
//aa.Id += 10000; // debug
if (aa.GeneSymbol != null)
{
aa.GeneSymbol = aa.GeneSymbol.ToUpper(); // be sure key match cols are upper case
}
if (aa.GeneFamily != null)
{
aa.GeneFamily = aa.GeneFamily.ToUpper();
}
if (aa.GeneFamilyTargetSymbol != null)
{
aa.GeneFamilyTargetSymbol = aa.GeneFamilyTargetSymbol.ToUpper();
}
if (aa.ResultTypeConcType != null)
{
aa.ResultTypeConcType = aa.ResultTypeConcType.ToUpper();
}
if (aa.AssayType != null)
{
aa.AssayType = aa.AssayType.ToUpper();
}
if (aa.AssayMode != null)
{
aa.AssayMode = aa.AssayMode.ToUpper();
}
AssayAttributesDao.InsertCommonAssayAttributes(aa, dao);
if (TimeOfDay.Milliseconds() - t0 > 1000)
{
//Progress.Show("Inserting new data " + (ri + 1) + "/" + rows.Count + "...");
t0 = TimeOfDay.Milliseconds();
}
}
dao.Commit();
dao.Dispose();
copyUpdateMsg = UpdateCmnAssyAtrbtsCopies();
}
string response =
"----------------------------------\r\n" +
"Assays processed: " + assaysProcessed + "\r\n" +
"Assays with processing errors: " + assaysWithProcessingErrors + "\r\n" +
"Rows inserted: " + dbRows.Count + "\r\n" +
copyUpdateMsg +
"----------------------------------\r\n" +
"Assays with CRC only: " + assaysWithCrcOnly + "\r\n" +
"Assays with SP only: " + assaysWithSpOnly + "\r\n" +
"Assays with CRC and SP: " + assaysWithCrcAndSp + "\r\n" +
"Assays with no CRC or SP: " + assaysWithNoCrcSP + "\r\n" +
"Assays with non CRC/SP key types: " + assaysWithOtherTypes + "\r\n" +
"Assays with no key types: " + assaysWithNoKeyTypes + "\r\n" +
"----------------------------------\r\n" +
"Assays with targets defined: " + assaysWithTargets + "\r\n" +
"Assays with genes defined: " + assaysWithGenes + "\r\n" +
"Assays with gene map coordinates: " + assaysWithGeneCoords + "\r\n" +
"----------------------------------\r\n" +
//"CRC Assays: " + CrcAssayCnt["ASSAY"] + "\r\n" +
//"SP Assays: " + SpAssayCnt["ASSAY"] + "\r\n" +
//"??? Assays: " + OtherAssayCnt["ASSAY"] + "\r\n" +
"----------------------------------";
LogMessage("\r\n" + response);
UAL.Progress.Hide();
return(response);
}
/// <summary>
/// Sync the Mobius CorpMoltable replicate used to retrieve Smiles
/// Syntax: UpdateCorpDbMoltableMx [ ByDateRange | ByCorpIdRange | LoadMissing | <singleCorpId>]
/// </summary>
/// <returns></returns>
public static string UpdateCorpDbMoltableMx(
string args)
{
DateTime moleculeDateTime = DateTime.MinValue;
double mw;
string msg = "", sql = "", maxCorpIdSql, mf = "", chime = "", smiles = "", checkPointDate = "", helm = "", sequence = "", svg = "";
object[][] pva = null;
int pvaCount = 0, CorpId, lowCorpId = 0, highCorpId = 0, srcMaxCorpId = 0;
int SelectChunkSize = 20; // small chunks
int InsertBufferSize = 10;
//int SelectChunkSize = 100000; // big chunks
//int InsertBufferSize = 1000;
// Select data from corp_moltable by CorpId range
const string SelectByCorpIdRange = @"
SELECT
m.corp_nbr,
chime(m.ctab),
m.molformula,
m.molweight,
null molsmiles,
s.helm_txt,
s.sequence_txt,
m.molecule_date
FROM
corp_owner.corp_moltable m,
corp_owner.corp_substance s
where
m.corp_nbr > 0
and s.corp_nbr = m.corp_nbr
and (s.status_code is null or s.status_code = 'A')
ORDER BY corp_nbr";
// Select data from corp_moltable by date range comparing to corp_moltable_mx
const string SelectByDateRange = @"
select
m.corp_nbr,
chime(m.ctab),
m.molformula,
m.molweight,
null molsmiles,
s.helm_txt,
s.sequence_txt,
m.molecule_date,
m2.molecule_date
from
corp_owner.corp_moltable m,
corp_owner.corp_substance s,
corp_moltable_mx m2
where
m.molecule_date > to_date('1-jan-1900 000000','DD-MON-YYYY HH24MISS')
and s.corp_nbr = M.CORP_NBR
and (s.status_code is null or s.status_code = 'A')
and m2.corp_nbr (+) = m.corp_nbr
and m2.molecule_date (+) != m.molecule_date
order by m.molecule_date" ;
// Select for missing smiles strings, ex: Update CorpDbMoltableMx LoadMissing mx.molecule_date > '1-jan-2014'
const string SelectMissingSmilesFix = @"
select /* check for CorpIds in corp_moltable not in corp_moltable_mx */
corp_nbr,
chime(ctab),
molformula,
molweight,
null molsmiles,
helm_txt,
sequence_txt,
molecule_date
from
(
select
m.*,
s.helm_txt,
s.sequence_txt,
mx.molsmiles molsmiles_mx
from
corp_owner.corp_moltable m,
corp_owner.corp_substance s,
corp_moltable_mx mx
where
s.corp_nbr = M.CORP_NBR
and (s.status_code is null or s.status_code = 'A')
and mx.corp_nbr (+) = m.corp_nbr
and 1=1 /* condition to substitute */
) m
where molsmiles_mx is null /* extra condition */
order by corp_nbr" ;
// Insert missing helm info
const string SelectMissingHelmFix = @"
select /* check for CorpIds in corp_moltable not in corp_moltable_mx */
corp_nbr,
chime(ctab),
molformula,
molweight,
null molsmiles,
helm_txt,
sequence_txt,
molecule_date
from
(
select
m.*,
s.helm_txt,
s.sequence_txt,
mx.molsmiles molsmiles_mx
from
corp_owner.corp_moltable m,
corp_owner.corp_substance s,
corp_moltable_mx mx
where
s.corp_nbr = M.CORP_NBR
and (s.status_code is null or s.status_code = 'A')
and mx.corp_nbr (+) = m.corp_nbr
and 1=1 /* condition to substitute */
) m
where length(helm_txt) > 0 /* extra condition */
order by corp_nbr" ;
// Secondary "large" structure table (~5k mols)
const string SelectLargeMols = @"
select
corp_nbr,
to_clob(molstructure),
to_clob(molformula),
molweight,
molsmiles,
null helm_txt,
null sequence_txt,
molecule_date
from
(select
corp_srl_nbr corp_nbr,
'CompoundId=' || corp_srl_nbr molstructure,
null ctab,
mlclr_frml_txt molformula,
mlclr_wgt molweight,
null molsmiles,
null molecule_date
from rdm_owner.rdm_sbstnc
where rdw_src_cd = 'LRG'" ;
// Insert statement
const string InsertSql = @"
insert into mbs_owner.corp_moltable_mx (
corp_nbr,
molstructure,
molformula,
molweight,
molsmiles,
molecule_date)
values (:0, :1, :2, :3, :4, :5)" ;
// Build select sql
bool byDateRange = false, byCorpIdRange = false, missingFix = true, deleteExisting = true;
string missingFixCriteria = "";
if (Lex.IsUndefined(args) || Lex.Eq(args, "ByDateRange"))
{
byDateRange = true;
}
else if (Lex.Eq(args, "ByCorpIdRange"))
{
byCorpIdRange = true;
Progress.Show("Getting range of CorpIds to insert...");
maxCorpIdSql = "select max(corp_nbr) from corp_owner.corp_moltable"; // get highest CorpId in source db
srcMaxCorpId = SelectSingleValueDao.SelectInt(maxCorpIdSql);
if (srcMaxCorpId < 0)
{
srcMaxCorpId = 0;
}
maxCorpIdSql = "select max(corp_nbr) from mbs_owner.corp_moltable_mx"; // get highest CorpId in dest db
highCorpId = SelectSingleValueDao.SelectInt(maxCorpIdSql);
if (highCorpId < 0)
{
highCorpId = 0;
}
}
else if (Lex.StartsWith(args, "LoadMissing"))
{
missingFix = true;
if (args.Contains(" "))
{
missingFixCriteria = args.Substring(10).Trim();
}
}
else if (int.TryParse(args, out srcMaxCorpId)) // single CorpId
{
byCorpIdRange = true;
highCorpId = srcMaxCorpId - 1; // say 1 less is the max we have
}
else
{
return("Syntax: UpdateCorpDbMoltableMx [ ByDateRange | ByCorpIdRange | LoadMissing | <singleCorpId>]");
}
Log("UpdateCorpDbMoltableMx started: " + args);
int readCount = 0, insCount = 0, insertCount = 0, updateCount = 0, undefinedStructures = 0, smilesSuccess = 0, smilesFails = 0, helmStructures = 0;
List <string> CorpIdList = new List <string>();
for (int chunk = 1; ; chunk++) // loop over chunks
{
if (byDateRange) // single chunk
{
if (chunk > 1)
{
break; // break 2nd time through
}
checkPointDate = UserObjectDao.GetUserParameter("MOBIUS", "UpdateCorpDbMoltableMxCheckpointDate", "01-sep-2013 000000");
//UserObjectDao.SetUserParameter("MOBIUS", "UpdateCorpDbMoltableMxCheckpointDate", checkPointDate);
sql = Lex.Replace(SelectByDateRange, "1-jan-1900 000000", checkPointDate);
msg = "Reading where date >= " + checkPointDate;
}
else if (byCorpIdRange) // by CorpId range
{
if (highCorpId >= srcMaxCorpId)
{
break; // done
}
lowCorpId = highCorpId + 1; // start of next chunk
highCorpId = lowCorpId + SelectChunkSize;
if (highCorpId >= srcMaxCorpId)
{
highCorpId = srcMaxCorpId;
}
sql = Lex.Replace(SelectByCorpIdRange, "corp_nbr > 0", "corp_nbr between " + lowCorpId + " and " + highCorpId);
msg = "Reading: " + lowCorpId + " to " + highCorpId + ", Reads: " + readCount + ", Inserts: " + insertCount;
}
else if (missingFix)
{
if (chunk > 1)
{
break; // break 2nd time through
}
sql = SelectMissingHelmFix;
if (Lex.IsDefined(missingFixCriteria)) // substitute any criteria
{
sql = Lex.Replace(sql, "1=1", missingFixCriteria);
}
msg = "Fixing missing data";
}
Progress.Show(msg);
DbCommandMx readCmd = new DbCommandMx();
readCmd.MxConn = DbConnectionMx.GetConnection("prd123");
readCmd.PrepareUsingDefinedConnection(sql, null);
DbDataReader rdr = readCmd.ExecuteReader();
DbCommandMx insertCmd = new DbCommandMx();
OracleDbType[] pta = new OracleDbType[6];
pta[0] = OracleDbType.Int32; // corp_nbr
pta[1] = OracleDbType.Clob; // molstructure
pta[2] = OracleDbType.Clob; // molformula
pta[3] = OracleDbType.Double; // molweight
pta[4] = OracleDbType.Clob; // smiles
pta[5] = OracleDbType.Date; // molecule_date
insertCmd.Prepare(InsertSql, pta);
insertCmd.BeginTransaction(); // be sure we have a transaction going
pva = DbCommandMx.NewObjectArrayArray(6, InsertBufferSize); // alloc insert row array
object[] vo = new object[6];
while (true)
{
bool readOk = rdr.Read();
if (readOk)
{
rdr.GetValues(vo);
CorpId = readCmd.GetInt(0); // corp_nbr
vo[0] = CorpId;
CorpIdList.Add(CorpId.ToString());
if (!readCmd.IsNull(1)) // molstructure
{
chime = readCmd.GetClob(1);
chime = OracleMx.ClearStringIfExceedsMaxStringSize(chime);
vo[1] = chime;
}
else
{
chime = "";
}
if (!readCmd.IsNull(2)) // molformula
{
mf = readCmd.GetClob(2);
mf = OracleMx.ClearStringIfExceedsMaxStringSize(mf);
vo[2] = mf;
}
if (!readCmd.IsNull(3)) // molweight
{
mw = readCmd.GetDouble(3);
vo[3] = mw;
}
if (Lex.IsDefined(chime)) // molsmiles - calculate from chime string
{
MoleculeMx cs = new MoleculeMx(MoleculeFormat.Chime, chime);
if (cs.AtomCount > 1) // need more than one atom
{
MoleculeMx cs2 = cs.ConvertTo(MoleculeFormat.Smiles);
smiles = cs2.GetSmilesString();
if (Lex.IsDefined(smiles))
{
smilesSuccess++;
}
else
{
Log("Smiles conversion failure for CorpId: " + CorpId);
smilesFails++;
}
smiles = OracleMx.ClearStringIfExceedsMaxStringSize(smiles);
vo[4] = smiles;
}
else
{
undefinedStructures++;
}
}
else
{
undefinedStructures++;
}
if (!readCmd.IsNull(5))
{
helm = readCmd.GetClob(5);
if (Lex.IsDefined(helm))
{
svg = HelmControl.GetSvg(helm);
vo[1] = SvgUtil.CompressSvgString(svg); // store compressed svg in molstructure column for now
helmStructures++;
}
}
if (!readCmd.IsNull(6))
{
sequence = readCmd.GetClob(6);
if (Lex.IsDefined(sequence))
{
// nothing yet
}
}
moleculeDateTime = DateTime.MinValue;
if (!readCmd.IsNull(7)) // molecule_date
{
moleculeDateTime = readCmd.GetDateTime(7);
vo[5] = moleculeDateTime;
}
for (int pi = 0; pi < 6; pi++) // invert for insert
{
pva[pi][pvaCount] = vo[pi];
}
if (Debug)
{
msg = String.Format("CorpId: {0}, mf: {1}, chime: {2}, smiles: {3}", CorpId.ToString(), mf.Length, chime.Length, smiles.Length);
Log(msg);
}
pvaCount++;
}
if (pvaCount >= InsertBufferSize || (!readOk && pvaCount > 0)) // write if buffer full or at end
{
try
{
if (deleteExisting)
{
int delCount = DoDeletes(CorpIdList);
updateCount += delCount; // count deletes as updates
insertCount -= delCount; // subtract from inserts
}
CorpIdList.Clear();
insCount = insertCmd.ExecuteArrayNonReader(pva, ref pvaCount);
insertCmd.Commit();
insertCmd.BeginTransaction();
insertCount += insCount;
}
catch (Exception ex)
{
throw new Exception(ex.Message, ex);
}
if (byDateRange)
{
string checkPointDate2 = String.Format("{0:dd-MMM-yyyy HHmmss}", moleculeDateTime); // format date time that will work with oracle
UserObjectDao.SetUserParameter("MOBIUS", "UpdateCorpDbMoltableMxCheckpointDate", checkPointDate2);
msg = "Processing where date >= " + checkPointDate + ", Reads: " + readCount + ", Inserts: " + insertCount + ", Updates: " + updateCount;
}
else if (byCorpIdRange) // CorpId range
{
msg = "Processing: " + lowCorpId + " to " + highCorpId + ", Reads: " + readCount + ", Inserts: " + insertCount;
}
else if (missingFix)
{
msg = "Fixing missing smiles, Updates: " + updateCount;
}
msg += String.Format(", Undefined structures: {0} , Smiles failures: {1}, Helms: {2}", undefinedStructures, smilesFails, helmStructures);
Progress.Show(msg);
}
if (!readOk)
{
break;
}
readCount++;
}
readCmd.Dispose();
insertCmd.Dispose();
} // end for select chunk
msg = "UpdateCorpDbMoltableMx - Inserts: " + insertCount + ", Updates: " + updateCount;
msg += String.Format(", Undefined structures: {0} , Smiles failures: {1}, Helms: {2}", undefinedStructures, smilesFails, helmStructures);
Log(msg);
return(msg);
}
/// <summary>
/// Load data into PubChem database
/// PubChem assay data files are downloaded from the PubChem site:
/// http://pubchem.ncbi.nlm.nih.gov/ using a program like SmartFTP.
/// The files are in GNU Zip (.gz) format and can be unzipped with
/// the following gzip commands:
/// c:\gzip\gzip -d c:\pubchem\bioassay\csv\description\*.gz
/// c:\gzip\gzip -d c:\pubchem\bioassay\csv\data\*.gz
/// After downloading and decompression this method can be called on the files.
/// </summary>
/// <param name="args"></param>
/// <returns></returns>
public static string LoadData(
string aid)
{
int recCount = 0;
string mtName = "PubChem_aid_" + aid;
MetaTable mt = MetaTableCollection.Get(mtName);
if (mt == null)
{
return("Failed to get metatable");
}
// if (Math.Sqrt(4) == 2) goto UpdateCids;
string fileName = PubChemAssayDirectory + @"\CSV\Data\" + aid + ".csv";
StreamReader sr;
try { sr = new StreamReader(fileName); }
catch (Exception ex) { return("File not found: " + fileName); }
string header = sr.ReadLine(); // read headers line
List <string> headers = Csv.SplitCsvString(header);
int cidIdx = -1;
for (cidIdx = 0; cidIdx < headers.Count; cidIdx++)
{
if (headers[cidIdx].ToUpper() == "PUBCHEM_CID")
{
break;
}
}
if (cidIdx >= headers.Count)
{
sr.Close();
return("PUBCHEM_CID column not found in data headers");
}
Dictionary <string, MetaColumn> mcd = new Dictionary <string, MetaColumn>();
foreach (MetaColumn mc2 in mt.MetaColumns)
{
mcd[mc2.Name.ToUpper()] = mc2; // build dict for quick metacolumn lookup
}
DbConnectionMx conn = DbConnectionMx.MapSqlToConnection(ref PubChemWarehouseTable);
conn.BeginTransaction(); // do multiple updates per transaction
GenericDwDao dao = new GenericDwDao(
PubChemWarehouseTable, // table for results
PubChemWarehouseSeq); // sequence to use
dao.BufferInserts(true); // buffer inserts for better speed
SequenceDao.SetCacheSize(PubChemWarehouseSeq, 100); // number of ids to cache locally from sequence
//string progressMsg = "Deleting existing data...";
int i1 = dao.DeleteTable(Int32.Parse(mt.TableFilterValues[0]), true);
//if (Progress.CancelRequested())
//{
// dao.Dispose();
// return "Cancelled during data delete";
//}
//Progress.Show("Loading file...");
recCount = 0;
int t1 = 0;
while (true)
{
int t2 = TimeOfDay.Milliseconds();
if (t2 - t1 > 1000)
{
if (Progress.CancelRequested)
{
dao.ExecuteBufferedInserts();
conn.Commit();
conn.Close();
sr.Close();
Progress.Hide();
return(recCount.ToString() + " rows loaded");
}
Progress.Show("Loading file (" + recCount.ToString() + ") ...");
t1 = t2;
}
string rec = sr.ReadLine();
if (rec == null)
{
break;
}
List <string> vals = Csv.SplitCsvString(rec);
int cid;
try { cid = Int32.Parse(vals[cidIdx]); } // get compound id
catch (Exception ex)
{
string txtCid = vals[cidIdx];
if (txtCid == null)
{
txtCid = "";
}
DebugLog.Message("Load PubChem bad CID " + txtCid + ", AID = " + aid);
continue;
}
long rslt_grp_id = dao.GetNextIdLong(); // id to hold row together
for (int vi = 0; vi < vals.Count; vi++)
{
string s = vals[vi];
if (s == "")
{
continue;
}
string[] sa = rec.Split(',');
if (vi >= headers.Count)
{
continue;
}
string mcName = headers[vi].ToUpper();
if (mcName.Length > 26)
{
mcName = mcName.Substring(0, 26); // limit length to 26
}
if (mcName == "PUBCHEM_CID")
{
continue;
}
if (Lex.IsInteger(mcName))
{
mcName = "R_" + mcName; // result number
}
MetaColumn mc = mcd[mcName];
if (mc == null)
{
continue;
}
AnnotationVo vo = new AnnotationVo();
vo.rslt_grp_id = rslt_grp_id;
if (mc.DataType == MetaColumnType.String)
{
vo.rslt_val_txt = s;
}
else if (mc.DataType == MetaColumnType.Number || mc.DataType == MetaColumnType.Integer)
{
try
{
vo.rslt_val_nbr = Convert.ToDouble(s);
}
catch (Exception e) { continue; } // just continue if bad
}
else if (mc.DataType == MetaColumnType.Date)
{
s = DateTimeMx.Normalize(s);
if (s == null)
{
continue;
}
vo.rslt_val_dt = DateTimeMx.NormalizedToDateTime(s);
}
else if (mc.Name == "PUBCHEM_ACTIVITY_OUTCOME") // activity outcome is a dict value stored as an integer
{
try
{
vo.rslt_val_nbr = Convert.ToInt32(s);
}
catch (Exception e) { continue; } // just continue if bad
}
else if (mc.DataType == MetaColumnType.Hyperlink ||
mc.DataType == MetaColumnType.DictionaryId)
{
vo.rslt_val_txt = s;
}
else
{
continue;
}
vo.ext_cmpnd_id_nbr = cid;
vo.ext_cmpnd_id_txt = cid.ToString();
vo.mthd_vrsn_id = Int32.Parse(mt.TableFilterValues[0]);
vo.rslt_typ_id = Int32.Parse(mc.PivotValues[0]);
vo.chng_op_cd = "I";
vo.chng_usr_id = Security.UserInfo.UserName;
dao.Insert(vo);
} // end of field loop
recCount++;
if (recCount % 100 == 0)
{ // commit after group of updates
dao.ExecuteBufferedInserts();
conn.Commit();
conn.BeginTransaction(); // do multiple updates per transaction
}
} // end of record loop
dao.ExecuteBufferedInserts();
conn.Commit();
conn.Close();
dao.Dispose();
sr.Close();
//UpdateCids: // Add any missing CIDs under method 1000000
Progress.Show("Updating CID table...");
string sql =
"INSERT INTO " + PubChemWarehouseTable + "(ext_cmpnd_id_nbr,rslt_id,mthd_vrsn_id,rslt_typ_id,rslt_grp_id) " +
"SELECT ext_cmpnd_id_nbr, " + PubChemWarehouseSeq + ".NEXTVAL,1000000,0,0 " +
"FROM ( " +
"SELECT UNIQUE ext_cmpnd_id_nbr " +
"FROM " + PubChemWarehouseTable + " r1 " +
"WHERE mthd_vrsn_id = " + aid + " " +
"AND NOT EXISTS ( " +
" SELECT * " +
"FROM " + PubChemWarehouseTable + " r2 " +
"WHERE mthd_vrsn_id = 1000000 " +
"AND r2.ext_cmpnd_id_nbr = r1.ext_cmpnd_id_nbr) " +
"and rownum <= 10000)";
DbCommandMx drd = new DbCommandMx();
drd.Prepare(sql);
drd.BeginTransaction();
int newCids = 0;
while (true)
{
int addedCids = drd.ExecuteNonReader();
if (addedCids == 0)
{
break;
}
newCids += addedCids;
drd.Commit();
drd.BeginTransaction(); // do multiple updates per transaction
Progress.Show("Updating CID table (" + newCids.ToString() + ")...");
}
drd.Dispose();
Progress.Hide();
return(recCount.ToString() + " rows loaded for AID " + aid + " plus " + newCids.ToString() + " new CIDs");
}