Bits *maskQuerySeq(struct dnaSeq *seq, boolean isProt,
boolean maskQuery, boolean lcMask)
/* Massage query sequence a bit, converting it to correct
* case (upper for protein/lower for DNA) and optionally
* returning upper/lower case info , and trimming poly A. */
{
Bits *qMaskBits = NULL;
verbose(2, "%s\n", seq->name);
if (isProt)
faToProtein(seq->dna, seq->size);
else
{
if (maskQuery)
{
if (lcMask)
toggleCase(seq->dna, seq->size);
qMaskBits = maskFromUpperCaseSeq(seq);
}
faToDna(seq->dna, seq->size);
}
if (seq->size > qWarnSize)
{
warn("Query sequence %s has size %d, it might take a while.",
seq->name, seq->size);
}
return qMaskBits;
}
void faToTwoBit(char *inFiles[], int inFileCount, char *outFile)
/* Convert inFiles in fasta format to outfile in 2 bit
* format. */
{
struct twoBit *twoBitList = NULL, *twoBit;
int i;
struct hash *uniqHash = newHash(18);
FILE *f;
for (i=0; i<inFileCount; ++i)
{
char *fileName = inFiles[i];
struct lineFile *lf = lineFileOpen(fileName, TRUE);
struct dnaSeq seq;
ZeroVar(&seq);
while (faMixedSpeedReadNext(lf, &seq.dna, &seq.size, &seq.name))
{
if (seq.size == 0)
{
warn("Skipping item %s which has no sequence.\n",seq.name);
continue;
}
/* strip off version number */
if (stripVersion)
{
char *sp = NULL;
sp = strchr(seq.name,'.');
if (sp != NULL)
*sp = '\0';
}
if (hashLookup(uniqHash, seq.name))
{
if (!ignoreDups)
errAbort("Duplicate sequence name %s", seq.name);
else
continue;
}
hashAdd(uniqHash, seq.name, NULL);
if (noMask)
faToDna(seq.dna, seq.size);
else
unknownToN(seq.dna, seq.size);
twoBit = twoBitFromDnaSeq(&seq, !noMask);
slAddHead(&twoBitList, twoBit);
}
lineFileClose(&lf);
}
slReverse(&twoBitList);
f = mustOpen(outFile, "wb");
twoBitWriteHeader(twoBitList, f);
for (twoBit = twoBitList; twoBit != NULL; twoBit = twoBit->next)
{
twoBitWriteOne(twoBit, f);
}
carefulClose(&f);
}
boolean faSomeSpeedReadNext(struct lineFile *lf, DNA **retDna, int *retSize, char **retName, boolean isDna)
/* Read in DNA or Peptide FA record. */
{
char *poly;
int size;
if (!faMixedSpeedReadNext(lf, retDna, retSize, retName))
return FALSE;
size = *retSize;
poly = *retDna;
if (isDna)
faToDna(poly, size);
else
faToProtein(poly, size);
return TRUE;
}
