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);
}
int TwoBitRunner::faToTwoBit(std::map<std::string, std::string> inputCommands) {
TwoBitSetUp setUp(inputCommands);
std::string inputFilename = "";
std::string outFilename = "";
bool overWrite = false;
bool trimNameAtWhitepsace = false;
setUp.setOption(inputFilename, "--in,-i", "Input fasta filename, can be several files seperated by commas", true);
setUp.setOption(outFilename, "--out,-o",
"Name of an output file", true);
setUp.setOption(overWrite, "--overWrite",
"Whether to overwrite the file if one is given by --out");
setUp.setOption(overWrite, "--overWrite",
"Whether to overwrite the file if one is given by --out");
setUp.setOption(trimNameAtWhitepsace, "--trimNameAtWhitepsace",
"Whether to trim the names of the fasta records at the first whitespace");
setUp.finishSetUp(std::cout);
cppprogutils::appendAsNeeded(outFilename, ".2bit");
std::ofstream out;
//check if output file exists
if (!overWrite && cppprogutils::fexists(outFilename)) {
throw Exception(__PRETTY_FUNCTION__,
"File " + outFilename
+ " already exists, use --overWrite to over write");
}
//read in seqs
std::vector<std::unique_ptr<FastaRecord>> seqs;
auto toks = cppprogutils::tokenizeString(inputFilename, ",");
for(const auto & fName : toks){
std::ifstream in(fName);
std::unique_ptr<FastaRecord> seq;
while (readNextFasta(in, seq, trimNameAtWhitepsace)) {
seqs.emplace_back(std::move(seq));
}
}
out.open(outFilename, std::ios::binary | std::ios::out);
//write out header
twoBitWriteHeader(seqs, out);
//write out sequences
for (const auto & seq : seqs) {
seq->twoBitWriteOne(out);
}
return 0;
}
void twoBitMask(char *inName, char *maskName, char *outName)
/* twoBitMask - apply masking to a .2bit file, creating a new .2bit file. */
{
struct hash *tbHash = hashNew(20);
struct hash *bitmapHash = hashNew(20);
struct twoBit *twoBitList = NULL;
struct twoBit *twoBit = NULL;
FILE *f = NULL;
if (! twoBitIsFile(inName))
{
if (twoBitIsSpec(inName))
errAbort("Sorry, this works only on whole .2bit files, not specs.");
else
errAbort("Input %s does not look like a proper .2bit file.", inName);
}
twoBitList = slurpInput(inName, tbHash, bitmapHash);
/* Read mask data into bitmapHash, store it in twoBits: */
if ((type && endsWith(type, "bed")) || endsWith(maskName, ".bed"))
maskWithBed(maskName, tbHash, bitmapHash);
else if ((type && endsWith(type, "out")) || endsWith(maskName, ".out"))
maskWithOut(maskName, tbHash, bitmapHash);
else
errAbort("Sorry, maskFile must end in \".bed\" or \".out\".");
/* Create a new .2bit file, write it out from twoBits. */
f = mustOpen(outName, "wb");
twoBitWriteHeader(twoBitList, f);
for (twoBit = twoBitList; twoBit != NULL; twoBit = twoBit->next)
{
twoBitWriteOne(twoBit, f);
}
carefulClose(&f);
/* Don't bother freeing twoBitList and hashes here -- just exit. */
}
