void aNotB(char *aFile, char *bFile, char *outFile)
/* aNotB - List symbols that are in a but not b. */
{
struct hash *bHash = hashFirstWord(bFile);
struct lineFile *lf = lineFileOpen(aFile, TRUE);
FILE *f = mustOpen(outFile, "w");
char *row[1];
while (lineFileRow(lf, row))
{
if (!hashLookup(bHash, row[0]))
fprintf(f, "%s\n", row[0]);
}
carefulClose(&f);
}
void hgSgdPep(char *inName, char *outName, char *symbolName)
/* hgSgdPep - Parse yeast protein fasta files into format we can load. */
{
struct lineFile *lf = lineFileOpen(inName, TRUE);
FILE *f = mustOpen(outName, "w");
FILE *symF = mustOpen(symbolName, "w");
boolean inPep = FALSE;
char *line;
char *pattern = "ORFP:";
int patternSize = strlen(pattern);
struct hash *restrictHash = NULL;
if (optionExists("restrict"))
restrictHash = hashFirstWord(optionVal("restrict", NULL));
while (lineFileNext(lf, &line, NULL))
{
if (line[0] == '>')
{
char *orf, *gene;
inPep = FALSE;
if ((line = stringIn(pattern, line)) != NULL)
{
line += patternSize;
orf = nextWord(&line);
gene = nextWord(&line);
if (orf == NULL)
errAbort("Expecting ORF name starting with Y line %d of %s",
lf->lineIx, lf->fileName);
if (gene != NULL)
fprintf(symF, "%s\t%s\n", orf, gene);
if (restrictHash == NULL || hashLookup(restrictHash, orf))
{
fprintf(f, ">%s\n", orf);
inPep = TRUE;
}
}
else
{
errAbort("No %s line %d of %s\n",
pattern, lf->lineIx, lf->fileName);
}
}
else if (inPep)
fprintf(f, "%s\n", line);
}
}
