void loadTagHash(struct hash *h, struct sageCounts *scList)
{
struct sageCounts *sc =NULL;
int count=0;
for(sc=scList;sc!=NULL;sc=sc->next)
{
if(count++ % 10000 == 0)
{
putTic();
}
hashAddUnique(h,sc->tag,sc);
}
printf("\tDone.\n");
}
void createSageSummary(char *tagList, char *tagCounts, char *outFile)
{
struct sage *sgList = NULL,*sg = NULL;
struct sageCounts *scList = NULL, *sc;
struct hash *scHash = newHash(16);
FILE *out = mustOpen(outFile,"w");
int i,count=0;
printf("Loading tagCounts.\n");
scList = sageCountsLoadAll(tagCounts);
printf("Loading SageTags.\n");
sgList = loadSageTags(tagList, scList->numExps);
printf("Inserting scList into hash, %d elements.\n",slCount(scList));
loadTagHash(scHash,scList);
printf("Calculating avgs and stdevs");
fflush(stdout);
for(sg=sgList; sg != NULL; sg=sg->next)
{
if(count++ %1000 == 0)
putTic();
scList = NULL;
for(i=0;i<sg->numTags;i++)
{
sc = hashFindVal(scHash,sg->tags[i]);
if(sc != NULL)
slSafeAddHead(&scList,sc);
}
for(i=0; i<sg->numExps; i++)
{
sg->exps[i] =i;
if(slCount(scList) > 3)
{
#ifdef BOGUS
int q = 2+2;
#endif
}
sg->meds[i] = scMedian(scList,i);
sg->aves[i] = scAverage(scList, i);
sg->stdevs[i] = scStdev(scList, i, sg->aves[i]);
}
sageTabOut(sg,out);
fflush(out);
}
carefulClose(&out);
printf("\tDone.\n");
}
void runSamples(char *goodFile, char *badFile, char *newDb, char *oldDb, int numToRun)
/* run a bunch of tests */
{
int i,j,k;
FILE *good = mustOpen(goodFile, "w");
FILE *bad = mustOpen(badFile, "w");
char *tmp = NULL;
int numGood=0, numBad=0, tooManyNs=0;
boolean success = FALSE;
struct dnaSeq *seq = NULL;
printf("Running Tests\t");
for(i=0;i<numToRun;i++)
{
struct coordConvRep *ccr = NULL;
struct coordConv *cc = NULL;
if(!(i%10)) putTic();
cc = getRandomCoord(oldDb);
seq = hDnaFromSeq(cc->chrom, cc->chromStart, cc->chromEnd, dnaLower);
if(!(strstr(seq->dna, "nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn")))
{
chrom = cc->chrom;
chromStart = cc->chromStart;
chromEnd = cc->chromEnd;
success = convertCoordinates(good, bad, printReport, printReport);
if(success)
numGood++;
else
numBad++;
}
else
{
tooManyNs++;
}
freeDnaSeq(&seq);
coordConvFree(&cc);
}
carefulClose(&good);
carefulClose(&bad);
printf("\tDone.\n");
printf("Out of %d attempts got %d 'succesfully converted' and %d 'had problems', %d had too many N's\n",
(numGood + numBad), numGood, numBad, tooManyNs);
printf("After checking got %d of %d correctly called and %d incorrectly called.\n",
hgTestCorrect, hgTestCorrect+hgTestWrong, hgTestWrong);
}
