src/hg/instinct/extractData/extractData.c 1.4
1.4 2010/01/14 07:12:28 jsanborn
updated
Index: src/hg/instinct/extractData/extractData.c
===================================================================
RCS file: /projects/compbio/cvsroot/kent/src/hg/instinct/extractData/extractData.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -b -B -U 1000000 -r1.3 -r1.4
--- src/hg/instinct/extractData/extractData.c 14 Jan 2010 05:45:49 -0000 1.3
+++ src/hg/instinct/extractData/extractData.c 14 Jan 2010 07:12:28 -0000 1.4
@@ -1,252 +1,264 @@
/* mapProbesToGenes - Will maps probes in BED format to overlapping gene(s). */
#include "common.h"
#include "linefile.h"
#include "hash.h"
#include "options.h"
#include "jksql.h"
#include "bed.h"
#include "genePred.h"
#include "hPrint.h"
#include "hdb.h"
#include "microarray.h"
#include "ra.h"
#include "featuresLib.h"
#include "hgHeatmapLib.h"
#include "cprob.h"
#include "hgStatsLib.h"
char *hgDb = "hg18";
char *genome = "Human";
void usage()
/* Explain usage and exit. */
{
errAbort(
"extractData \n"
- " extractData [OPTIONS] db table samplesgenes(s)\n"
+ " extractData [OPTIONS] db table sample_id genes(s)\n"
"options:\n"
" -median Output median value if multiple probes\n"
+ " -tcga Handle TCGA ids, keeps only first 16 chars in sample_id\n"
);
}
boolean median = FALSE;
+boolean isTCGA = FALSE;
static struct optionSpec options[] = {
{"median", OPTION_BOOLEAN},
+ {"tcga", OPTION_BOOLEAN},
{NULL, 0},
};
struct maGrouping *getMaGrouping(struct sqlConnection *hgConn, char *tableName)
{
/*microarray specific settings*/
struct trackDb *tdb = hMaybeTrackInfo(hgConn, tableName);
struct microarrayGroups *maGs = maGroupings("hg18", tableName);
trackDbFreeList(&tdb);
if (!maGs)
return NULL;
return maGs->allArrays;
}
struct hash *getSettings(char *tableName)
{
struct column *raList = getColumns(NULL, "datasets.ra", NULL);
struct column *col;
struct hash *settings = NULL;
for (col = raList; col; col = col->next)
{
if (!sameString(col->name, tableName))
continue;
settings = col->settings;
break;
}
if (!settings)
errAbort("Couldn't find datasets.ra listing for %s", tableName);
return settings;
}
struct geneAlias {
struct geneAlias *next;
char *gene;
struct slName *probes;
};
struct hash *getAliases(struct sqlConnection *hgConn, char *tableName,
struct slName *genes)
{
if (!hgConn || !tableName || !genes)
return NULL;
struct slName *sl;
struct dyString *dy = newDyString(100);
dyStringPrintf(dy,
"select * from %s where alias in (",
tableName);
for (sl = genes; sl; sl = sl->next)
{
dyStringPrintf(dy, "\"%s\"", sl->name);
if (sl->next)
dyStringPrintf(dy, ",");
}
dyStringPrintf(dy, ");");
char *query = dyStringCannibalize(&dy);
struct sqlResult *sr = sqlGetResult(hgConn, query);
struct geneAlias *ga, *gaList = NULL;
struct hash *gaHash = hashNew(0);
char **row;
while ((row = sqlNextRow(sr)) != NULL)
{
char *probe = cloneString(row[0]);
char *gene = cloneString(row[1]);
struct hashEl *el = hashLookup(gaHash, gene);
if (!el)
{
ga = AllocA(struct geneAlias);
ga->gene = cloneString(gene);
ga->probes = NULL;
slAddHead(&gaList, ga);
hashAdd(gaHash, gene, ga);
}
else
ga = el->val;
slNameAddHead(&ga->probes, probe);
}
sqlFreeResult(&sr);
return gaHash;
}
struct slDouble *getProbeData(struct sqlConnection *hgConn, int expId,
char *dataTable, struct slName *probes)
{
struct slName *sl;
struct dyString *dy = newDyString(100);
dyStringPrintf(dy,
"select * from %s where name in(",
dataTable);
for (sl = probes; sl; sl = sl->next)
{
dyStringPrintf(dy, "\"%s\"", sl->name);
if (sl->next)
dyStringPrintf(dy, ",");
}
dyStringPrintf(dy, ");");
char *query = dyStringCannibalize(&dy);
/* Get bed15 data from hg18 database */
struct sqlResult *sr = sqlGetResult(hgConn, query);
char **row = NULL;
struct slDouble *sdList = NULL;
while ((row = sqlNextRow(sr)) != NULL)
{
struct bed *nb = bedLoadN(row+1, 15);
if (expId >= nb->expCount)
continue;
struct slDouble *sd = slDoubleNew(nb->expScores[expId]);
slAddHead(&sdList, sd);
bedFree(&nb);
}
sqlFreeResult(&sr);
return sdList;
}
void extractData(char *tableName, char *sampleName, char *geneList)
{
if (!geneList)
return;
+char *sample = sampleName;
+if (isTCGA)
+ sample = cloneStringZ(sampleName, 16);
struct slName *genes = slNameListFromComma(geneList);
struct hashEl *el;
struct sqlConnection *hgConn = hAllocConnProfile("localDb", hgDb);
/* Set up datasets entry */
struct maGrouping *allA = getMaGrouping(hgConn, tableName);
if (!allA)
errAbort("Could not find maGrouping for %s!", tableName);
int i = 0, expId = -1;
for (; i < allA->size; i++)
{
- if (!sameString(allA->names[i], sampleName))
+ char *maName = allA->names[i];
+ if (isTCGA)
+ maName = cloneStringZ(allA->names[i], 16);
+
+ if (!sameString(maName, sample))
continue;
expId = allA->expIds[i];
break;
}
if (expId < 0)
- errAbort("Couldn't find sample %s in %s.\n", sampleName, tableName);
+ errAbort("Couldn't find sample %s in %s.\n", sample, tableName);
struct hash *settings = getSettings(tableName);
el = hashLookup(settings, "aliasTable");
if (!el)
errAbort("No aliasTable.\n");
char *aliasTable = cloneString(el->val);
struct hash *gaHash = getAliases(hgConn, aliasTable, genes);
struct hashCookie cookie = hashFirst(gaHash);
while ((el = hashNext(&cookie)) != NULL)
{
char *name = el->name;
struct geneAlias *ga = el->val;
struct slDouble *sd, *sdList = getProbeData(hgConn, expId, tableName, ga->probes);
if (!sdList)
continue;
printf("%s\t", name);
if (median)
{
double val = slDoubleMedian(sdList);
printf("%f\n", val);
}
else
{
struct dyString *dy = newDyString(100);
for (sd = sdList; sd; sd = sd->next)
{
dyStringPrintf(dy, "%f", sd->val);
if (sd->next)
dyStringPrintf(dy, ",");
}
char *val = dyStringCannibalize(&dy);
printf("%s\n", val);
}
}
hFreeConn(&hgConn);
}
int main(int argc, char *argv[])
/* Process command line. */
{
optionInit(&argc, argv, options);
if (argc != 4)
usage();
if (optionExists("median"))
median = TRUE;
+if (optionExists("tcga"))
+ isTCGA = TRUE;
extractData(argv[1], argv[2], argv[3]);
return 0;
}