void pslToBed(char *pslFile, char *bedFile, struct hash *cdsHash, bool doPosName)
/* pslToBed -- tranform a psl format file to a bed format file */
{
struct lineFile *pslLf = pslFileOpen(pslFile);
FILE *bedFh = mustOpen(bedFile, "w");
struct psl *psl;
while ((psl = pslNext(pslLf)) != NULL)
{
struct bed *bed = bedFromPsl(psl);
if (doPosName)
{
char *newName = needMem(512);
safef(newName, 512, "%s:%d-%d", psl->qName, psl->qStart, psl->qEnd);
freeMem(bed->name);
bed->name = newName;
}
if (cdsHash)
{
struct cds *cds = hashFindVal(cdsHash, psl->qName);
if (cds == NULL)
bed->thickStart = bed->thickEnd = bed->chromStart;
else
setThick(psl, bed, cds);
}
bedTabOutN(bed, 12, bedFh);
bedFree(&bed);
pslFree(&psl);
}
carefulClose(&bedFh);
lineFileClose(&pslLf);
}
struct bed *orthoBedFromPsl(struct sqlConnection *conn, char *db, char *orthoDb,
char *netTable, struct psl *psl)
/** Produce a bed on the orthologous genome from the original psl. */
{
struct bed *bed = NULL, *orthoBed = NULL;
int i;
bed = bedFromPsl(psl);
orthoBed = orthoBedFromBed(conn, db, orthoDb, netTable, bed);
bedFree(&bed);
return orthoBed;
}
struct bed *createBedsFromPsls(char *pslFile, int expCount)
/** creates a list of beds from a pslfile, allocates memory for
arrays as determined by expCount */
{
struct psl *pslList = NULL, *psl = NULL;
struct bed *bedList = NULL, *bed = NULL;
pslList = pslLoadAll(pslFile);
for(psl = pslList; psl != NULL; psl = psl->next)
{
bed = bedFromPsl(psl);
freez(&bed->name);
bed->name=parseNameFromHgc(psl->qName);
bed->score = 0;
bed->expCount = 0;
bed->expIds = needMem(sizeof(int)*expCount);
bed->expScores = needMem(sizeof(float)*expCount);
slAddHead(&bedList,bed);
}
slReverse(&bedList);
pslFreeList(&pslList);
return bedList;
}
void affyPslAndAtlasToBedNew(char *pslFile, char *atlasFile, char *bedOut,
char *expRecOut)
/** Main function that does all the work for new-style*/
{
struct lineFile *lf = lineFileOpen(atlasFile, TRUE);
char *line, *name;
int i, wordCount, expCount;
char **row;
double *data, median;
double invMedian, ratio, logRatio;
char *affyId;
struct hash *hash = newHash(17);
struct psl *psl;
struct bed *bed;
FILE *f = NULL;
int dataCount = 0, pslCount = 0, bedCount = 0;
int minExpVal = 20;
/* Open Atlas file and use first line to create experiment table. */
if (!lineFileNextReal(lf, &line))
errAbort("%s is empty", lf->fileName);
if (startsWith("Affy", line))
line += 4;
if (line[0] != '\t')
errAbort("%s doesn't seem to be a new format atlas file", lf->fileName);
expCount = lineToExp(line+1, expRecOut);
if (expCount <= 0)
errAbort("No experiments in %s it seems", lf->fileName);
warn("%d experiments\n", expCount);
f = mustOpen(bedOut, "w");
/* Build up a hash keyed by affyID with an int array of data
* for value. Do output in short case. */
AllocArray(row, expCount);
while (lineFileNextReal(lf, &line))
{
affyId = nextWord(&line);
wordCount = chopByWhite(line, row, expCount);
if (wordCount != expCount)
errAbort("Expecting %d data points, got %d line %d of %s",
expCount, wordCount, lf->lineIx, lf->fileName);
if (hashLookup(hash, affyId))
{
warn("Duplicate %s, skipping all but first.", affyId);
continue;
}
AllocArray(data, expCount);
for (i=0; i<expCount; ++i)
{
data[i] = atof(row[i]);
if (data[i] < minExpVal)
data[i] = minExpVal;
}
median = findPositiveMedian(data, expCount, minExpVal);
if (median >= 0)
{
invMedian = 1.0/median;
for (i=0; i<expCount; ++i)
{
double val = data[i];
val = safeLog2(invMedian*val);
data[i] = val;
}
if (shortOut)
shortDataOut(f, affyId, expCount, data);
else
hashAdd(hash, affyId, data);
}
data = NULL;
++dataCount;
}
lineFileClose(&lf);
warn("%d rows of expression data\n", dataCount);
/* Stream through psl file, converting it to bed with expression data. */
if (!shortOut)
{
lf = pslFileOpen(pslFile);
while ((psl = pslNext(lf)) != NULL)
{
++pslCount;
/* get probe id from sequence name */
name=parseNameFromHgc(psl->qName);
data = hashFindVal(hash, name);
if (data != NULL)
{
struct bed *bed = bedFromPsl(psl);
bed->expCount = expCount;
AllocArray(bed->expIds, expCount);
AllocArray(bed->expScores, expCount);
for (i=0; i<expCount; ++i)
{
bed->expScores[i] = data[i];
bed->expIds[i] = i;
}
bedTabOutN(bed, 15, f);
++bedCount;
bedFree(&bed);
}
pslFree(&psl);
}
warn("%d records in %s", pslCount, pslFile);
warn("%d records written to %s", bedCount, bedOut);
}
lineFileClose(&lf);
carefulClose(&f);
}
void outputBedsFromPsls(struct hash *pslHash,char *bedOutName, char *expRecordOutName,
char *affyFileName, char *expFileName)
/** For each set of entries in affyFile find matching psl and create a bed. */
{
struct bed *bed = NULL, *b=NULL;
struct psl *pslList = NULL, *psl = NULL;
struct hash *expHash = NULL;
int numExps = 0;
int expCount = 0;
int i =0;
char *probeSet = NULL;
char *row[4];
char key[128];
struct slName *expNames = NULL, *name = NULL;
FILE *bedOut = NULL;
FILE *expRecordOut = NULL;
char *toDiffFileName = optionVal("toDiffFile", NULL);
FILE *toDiffOut = NULL;
struct lineFile *lf = NULL;
fillInExpHash(expFileName, &expHash, &expNames, &expCount);
lf = lineFileOpen(affyFileName, TRUE);
bedOut = mustOpen(bedOutName, "w");
if(toDiffFileName != NULL)
toDiffOut = mustOpen(toDiffFileName, "w");
/* Loop through either adding experiments to beds or if new
probeset create bed from psl and start over. */
while(lineFileChopNextTab(lf, row, sizeof(row)))
{
/* Do we have to make a new bed? */
if(probeSet == NULL || differentWord(probeSet, row[0]))
{
occassionalDot();
numExps = 0;
/* If we have probeset print out the current beds. */
if(probeSet != NULL)
{
for(b = bed; b != NULL; b = b->next)
{
int avgCount = 0;
for(i = 0; i < b->expCount; i++)
if(b->expScores[i] != -10000)
avgCount++;
if(avgCount != 0 && b->score > 0)
b->score = log(b->score / avgCount) * 100;
else
b->score = 0;
bedTabOutN(b, 15, bedOut);
if(toDiffOut != NULL)
outputToDiffRecord(b, expNames, toDiffOut);
}
}
bedFreeList(&bed);
/* Lookup key in pslHash to find list of psl. */
safef(key, sizeof(key), "%s", row[0]);
pslList = hashFindVal(pslHash, key);
/* Can have multiple psls. */
for(psl = pslList; psl != NULL; psl = psl->next)
{
b = bedFromPsl(psl);
AllocArray(b->expIds, expCount );
AllocArray(b->expScores, expCount);
b->expCount = expCount;
initBedScores(b, expCount);
slAddHead(&bed, b);
}
}
if(bed != NULL)
{
/* Allocate larger arrays if necessary. */
if(numExps > expCount)
{
errAbort("Supposed to be %d experiments but probeset %s has at least %d",
expCount, bed->name, numExps);
}
for(b = bed; b != NULL; b = b->next)
{
int exp = hashIntVal(expHash, row[1]);
if(differentWord(row[3], "NaN"))
b->expScores[exp] = atof(row[3]);
if(differentWord(row[2], "NaN"))
b->score += atof(row[2]);
}
numExps++;
}
freez(&probeSet);
probeSet = cloneString(row[0]);
}
expRecordOut = mustOpen(expRecordOutName, "w");
i = 0;
for(name = expNames; name != NULL; name = name->next)
{
subChar(name->name, ',', '_');
subChar(name->name, ' ', '_');
fprintf(expRecordOut, "%d\t%s\tuclaExp\tuclaExp\tuclaExp\tuclaExp\t1\t%s,\n", i++, name->name, name->name);
}
hashFree(&expHash);
slFreeList(&expNames);
carefulClose(&expRecordOut);
carefulClose(&bedOut);
lineFileClose(&lf);
}
