struct encodeExp *encodeExpLoadAll(char *fileName)
/* Load all encodeExp from a whitespace-separated file.
* Dispose of this with encodeExpFreeList(). */
{
struct encodeExp *list = NULL, *el;
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[8];
while (lineFileRow(lf, row))
{
el = encodeExpLoad(row);
slAddHead(&list, el);
}
lineFileClose(&lf);
slReverse(&list);
return list;
}
struct encodeExp *encodeExpLoadAllByChar(char *fileName, char chopper)
/* Load all encodeExp from a chopper separated file.
* Dispose of this with encodeExpFreeList(). */
{
struct encodeExp *list = NULL, *el;
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[8];
while (lineFileNextCharRow(lf, chopper, row, ArraySize(row)))
{
el = encodeExpLoad(row);
slAddHead(&list, el);
}
lineFileClose(&lf);
slReverse(&list);
return list;
}
struct encodeExp *encodeExpLoadByQuery(struct sqlConnection *conn, char *query)
/* Load all encodeExp from table that satisfy the query given.
* Where query is of the form 'select * from example where something=something'
* or 'select example.* from example, anotherTable where example.something =
* anotherTable.something'.
* Dispose of this with encodeExpFreeList(). */
{
struct encodeExp *list = NULL, *el;
struct sqlResult *sr;
char **row;
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
el = encodeExpLoad(row);
slAddHead(&list, el);
}
slReverse(&list);
sqlFreeResult(&sr);
return list;
}
void encodeExpToTab(char *outExp, char *outSeries, char *outResults)
/* encodeExpToCvDb - Convert encode experiments table to a table more suitable for cvDb. */
{
struct hash *optHash = optionalFieldsHash();
struct mdbObj *mdbList = getMdbList(metaDbs, ArraySize(metaDbs));
struct hash *mdbHash = mdbHashKeyedByExpId(mdbList);
struct hash *seriesHash = hashNew(0);
struct series *seriesList = NULL;
verbose(1, "read %d mdb objects from %s in %d databases\n", mdbHash->elCount, metaTable,
(int)ArraySize(metaDbs));
struct sqlConnection *expDbConn = sqlConnect(expDb);
struct sqlConnection *cvDbConn = sqlConnect(cvDb);
char query[256];
sqlSafef(query, sizeof(query), "select * from %s", expTable);
struct sqlResult *sr = sqlGetResult(expDbConn, query);
FILE *f = mustOpen(outExp, "w");
char **row;
while ((row = sqlNextRow(sr)) != NULL)
{
/* Read in database structure. */
struct encodeExp *ee = encodeExpLoad(row);
/* Much of the data we're processing comes from lists of the form
* "a=aVal b=bVal c=cVal." We'll convert these to id's in the appropriate
* tables and store the IDs in the optCol array declared below. */
int optColCount = ArraySize(expOptionalFields);
int optCol[optColCount];
int i;
for (i=0; i<optColCount; ++i)
optCol[i] = 0;
/* Convert var=val string in encodeExp.expVars into list of slPairs, and loop through it. */
struct slPair *varList = slPairListFromString(ee->expVars, TRUE);
struct slPair *var;
for (var = varList; var != NULL; var = var->next)
{
/* Figure out name of table and the term within that table. */
char *table = var->name;
char *term = var->val;
if (sameString(table, "antibody")) // Deal with antibody special case
{
if (sameString(term, "Control") || sameString(term, "Input")
|| sameString(term, "RevXlinkChromatin") || sameString(term, "ripInput"))
{
table = "control";
}
}
/* If it looks like we have a valid table and term, store result in
* optCol array we'll output soon. */
struct hashEl *hel;
if ((hel = hashLookup(optHash, table)) != NULL)
{
int id = lookupId(cvDbConn, table, term);
if (id == 0)
{
warn("No id in cvDb for %s=%s\n", table, term);
continue;
}
int optColIx = ptToInt(hel->val);
optCol[optColIx] = id;
}
else
verbose(2, "%s %s ?\n", table, term);
}
/* Now we want to process metaDb, which has some info encodeExp does not. */
char *composite = NULL;
char ixAsString[16];
safef(ixAsString, sizeof(ixAsString), "%d", ee->ix);
struct mdbObj *mdb = hashFindVal(mdbHash, ixAsString);
if (mdb != NULL)
{
struct mdbVar *v;
for (v = mdb->vars; v != NULL; v = v->next)
{
/* Look up table and term and change table name if need be */
char *table = v->var;
char *term = v->val;
if (sameString(table, "antibody"))
table = "ab";
else if (sameString(table, "grant"))
table = "grantee";
/* Squirrel away the ever-important composite term for later. */
if (sameString("composite", table))
composite = term;
struct hashEl *hel;
if ((hel = hashLookup(optHash, table)) != NULL)
{
int optColIx = ptToInt(hel->val);
if (optCol[optColIx] == 0) // Only use mdb if encodeExp has no data.
{
int id = lookupId(cvDbConn, table, term);
optCol[optColIx] = id;
}
}
}
}
/* If we've got a composite, then make up a series record. */
if (composite != NULL)
{
assert(mdb != NULL);
struct series *series = hashFindVal(seriesHash, composite);
if (series == NULL)
{
series = seriesFromMdb(mdb, composite);
hashAdd(seriesHash, composite, series);
slAddHead(&seriesList, series);
}
}
if (ee->accession != NULL)
{
/* Write out required fields. Order of required fields
* here needs to follow order in expRequiredFields. */
fprintf(f, "%u", ee->ix);
fprintf(f, "\t%s", ee->updateTime);
fprintf(f, "\t%s", naForNull(composite));
fprintf(f, "\t%s", ee->accession);
fprintf(f, "\t%d", lookupId(cvDbConn, "organism", ee->organism));
fprintf(f, "\t%d", lookupId(cvDbConn, "lab", ee->lab));
fprintf(f, "\t%d", lookupId(cvDbConn, "dataType", ee->dataType));
fprintf(f, "\t%d", lookupId(cvDbConn, "cellType", ee->cellType));
/* Now write out optional fields. */
for (i=0; i<optColCount; ++i)
fprintf(f, "\t%d", optCol[i]);
/* End output record. */
fprintf(f, "\n");
}
}
/* Write out series list to a separate file. */
slReverse(&seriesList);
writeSeriesList(outSeries, seriesList);
/* Write out results to a separate file. */
writeMdbListAsResults(mdbList, outResults);
/* Clean up and go home. */
carefulClose(&f);
sqlFreeResult(&sr);
sqlDisconnect(&expDbConn);
sqlDisconnect(&cvDbConn);
}
