TwoMotif findMotifInStream(SpotterParams p, std::string streamDir,
length_t minMotifLen, length_t maxMotifLen) {
int numDims = 1;
Stream dataStream = readStreamFromDir(streamDir, numDims);
return findMotif(p, &dataStream.data[0], dataStream.data.size(),
minMotifLen, maxMotifLen);
}
int main(int argc, char *argv[])
/* Process command line. */
{
int i;
char *cp;
unsigned long long reversed;
size_t maxAlloc;
char asciiAlloc[32];
optionInit(&argc, argv, options);
if (argc < 2)
usage();
maxAlloc = 2100000000 *
(((sizeof(size_t)/4)*(sizeof(size_t)/4)*(sizeof(size_t)/4)));
sprintLongWithCommas(asciiAlloc, (long long) maxAlloc);
verbose(4, "#\tmaxAlloc: %s\n", asciiAlloc);
setMaxAlloc(maxAlloc);
/* produces: size_t is 4 == 2100000000 ~= 2^31 = 2Gb
* size_t is 8 = 16800000000 ~= 2^34 = 16 Gb
*/
dnaUtilOpen();
motif = optionVal("motif", NULL);
chr = optionVal("chr", NULL);
strand = optionVal("strand", NULL);
bedOutput = optionExists("bedOutput");
wigOutput = optionExists("wigOutput");
if (wigOutput)
bedOutput = FALSE;
else
bedOutput = TRUE;
if (chr)
verbose(2, "#\tprocessing chr: %s\n", chr);
if (strand)
verbose(2, "#\tprocessing strand: '%s'\n", strand);
if (motif)
verbose(2, "#\tsearching for motif: %s\n", motif);
else {
warn("ERROR: -motif string empty, please specify a motif\n");
usage();
}
verbose(2, "#\ttype output: %s\n", wigOutput ? "wiggle data" : "bed format");
verbose(2, "#\tspecified sequence: %s\n", argv[1]);
verbose(2, "#\tsizeof(motifVal): %d\n", (int)sizeof(motifVal));
if (strand)
{
if (! (sameString(strand,"+") | sameString(strand,"-")))
{
warn("ERROR: -strand specified ('%s') is not + or - ?\n", strand);
usage();
}
/* They are both on by default, turn off the one not specified */
if (sameString(strand,"-"))
doPlusStrand = FALSE;
if (sameString(strand,"+"))
doMinusStrand = FALSE;
}
motifLen = strlen(motif);
/* at two bits per character, size limit of motif is
* number of bits in motifVal / 2
*/
if (motifLen > (4*sizeof(motifVal))/2 )
{
warn("ERROR: motif string too long, limit %d\n", (4*(int)sizeof(motifVal))/2 );
usage();
}
cp = motif;
motifVal = 0;
complementVal = 0;
for (i = 0; i < motifLen; ++i)
{
switch (*cp)
{
case 'a':
case 'A':
motifVal = (motifVal << 2) | A_BASE_VAL;
complementVal = (complementVal << 2) | T_BASE_VAL;
break;
case 'c':
case 'C':
motifVal = (motifVal << 2) | C_BASE_VAL;
complementVal = (complementVal << 2) | G_BASE_VAL;
break;
case 'g':
case 'G':
motifVal = (motifVal << 2) | G_BASE_VAL;
complementVal = (complementVal << 2) | C_BASE_VAL;
break;
case 't':
case 'T':
motifVal = (motifVal << 2) | T_BASE_VAL;
complementVal = (complementVal << 2) | A_BASE_VAL;
break;
default:
warn(
"ERROR: character in motif: '%c' is not one of ACGT\n", *cp);
usage();
}
++cp;
}
reversed = 0;
for (i = 0; i < motifLen; ++i)
{
int base;
base = complementVal & 3;
reversed = (reversed << 2) | base;
complementVal >>= 2;
}
complementVal = reversed;
verbose(2, "#\tmotif numerical value: %llu (%#llx)\n", motifVal, motifVal);
verbose(2, "#\tcomplement numerical value: %llu (%#llx)\n", complementVal, complementVal);
if (motifLen < 5)
{
warn("ERROR: motif string must be more than 4 characters\n");
usage();
}
findMotif(argv[1]);
return 0;
}
TwoMotif findMotifInStream(SpotterParams p, Stream dataStream,
length_t minMotifLen, length_t maxMotifLen) {
return findMotif(p, &dataStream.data[0], dataStream.data.size(),
minMotifLen, maxMotifLen);
}
