Example #1
0
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);
}
Example #2
0
void hgGnfMicroarray(char *expTable, char *dataTable, char *atlasFile)
/** Main function that does all the work for new-style*/
{
struct lineFile *lf = lineFileOpen(atlasFile, TRUE);
char *line;
int i, wordCount, expCount;
char **row;
float *data;
char *affyId;
struct hash *hash = newHash(17);
FILE *f = NULL;
int dataCount = 0;

/* 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 (startsWith("Gene Name", line))
    line += 9;
if (line[0] != '\t')
    errAbort("%s doesn't seem to be a new format atlas file", lf->fileName);
expCount = lineToExpTable(line+1, expTable);
if (expCount <= 0)
    errAbort("No experiments in %s it seems", lf->fileName);
warn("%d experiments\n", expCount);

f = hgCreateTabFile(tabDir, dataTable);

AllocArray(row, expCount);
AllocArray(data, 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 (chopName != NULL)
        {
	char *e = stringIn(chopName, affyId);
	if (e != NULL)
	    *e = 0;
	}
    if (hashLookup(hash, affyId))
	{
        warn("Duplicate %s, skipping all but first.", affyId);
	continue;
	}
    for (i=0; i<expCount; ++i)
        {
        data[i] = sqlFloat(row[i]);
        }
    shortDataOut(f, affyId, expCount, data);
    ++dataCount;
    if (limit != 0 && dataCount >= limit)
        break;
    }
lineFileClose(&lf);

if (doLoad)
    {
    struct sqlConnection *conn = sqlConnect(database);
    expDataCreateTable(conn, dataTable);
    hgLoadTabFile(conn, tabDir, dataTable, &f);
    hgRemoveTabFile(tabDir, dataTable);
    sqlDisconnect(&conn);
    }
}