std::vector<char*> getRegions(const char * name){
if(!fexists(name)){
fprintf(stderr,"\t-> Problems opening file: %s\n",name);
exit(0);
}
const char* delims = " \t\n\r";
std::vector<char*> ret;
FILE *fp =getFILE(name,"r");
char buffer[fsize(name)+1];
if(fsize(name)!=fread(buffer,1,fsize(name),fp))
fprintf(stderr,"[%s] Problems reading %lu from: %s\n",__FUNCTION__,fsize(name),name);
buffer[fsize(name)]='\0';
char *tok = strtok(buffer,delims);
while(tok!=NULL){
if(tok[0]!='#'){
ret.push_back(strdup(tok));
}
tok = strtok(NULL,delims);
}
fprintf(stderr,"\t-> From regionsfile: %s we read %lu\n",name,ret.size());
fclose(fp);
return ret;
}
double *readDouble(const char*fname,int hint){
FILE *fp = getFILE(fname,"r");
char buf[fsize(fname)+1];
if(fsize(fname)!=fread(buf,sizeof(char),fsize(fname),fp)){
fprintf(stderr,"Problems reading file: %s\n will exit\n",fname);
exit(0);
}
buf[fsize(fname)]='\0';
std::vector<double> res;
res.push_back(atof(strtok(buf,"\t\n ")));
char *tok=NULL;
while((tok=strtok(NULL,"\t\n "))) {
//fprintf(stderr,"%s\n",tok);
res.push_back(atof(tok));
}
// fprintf(stderr,"size of prior=%lu\n",res.size());
if(hint!=res.size()){
fprintf(stderr,"problem with size of dimension of prior %d vs %lu\n",hint,res.size());
for(uint i=0;i<res.size();i++)
fprintf(stderr,"%d=%f\n",i,res[i]);
exit(0);
}
double *ret = new double[res.size()];
for(uint i=0;i<res.size();i++)
ret[i] = log(res[i]);
fclose(fp);
return ret;
}
/**
* DATE: 2010-9-15
* FUNCTION: open sfs file
* PARAMETER: outfiles : outfile name.
writeFr : whether writer frequence file.
doBay : whether do Bay.
doJoint : whether do Joint
* RETURN: OPENSFS_ERROR:if can not open file ,OPENSFS_SUCC : if can open file.
*/
int Files::OpenSfsfile(const string outfiles, const int writeFr, const int doBay, const int doJoint)
{
//generete the output filenames
if(!outfiles.c_str())
{
printf("Must supply -outfiles (-fai)\n");
return OPENSFS_ERROR;
}
fSFSall = outfiles + ".sfs";
fFreq = outfiles + ".frq";
fJoint = outfiles + ".bjoint";
//open the persite FILE streams
if(writeFr)
{
//freqfile.clear();
//freqfile.open(fFreq.c_str());
freqfile= getFILE(fFreq.c_str(),"w");
}
if(doBay)
{
//sfsfile.clear();
//sfsfile.open(fSFSall.c_str());
sfsfile = getFILE(fSFSall.c_str(),"w");
//sfsfile-open(fSFSall.c_str());
}
if(doJoint)
{
//jointSfsfile.clear();
//jointSfsfile.open(fJoint.c_str(), ios::binary);
jointSfsfile = getFILE(fJoint.c_str(),"w");
}
if ((writeFr && !freqfile) || (doBay && !sfsfile) || (doJoint && !jointSfsfile))
{
cerr << "\topen sfs file failed!" << endl;
return OPENSFS_ERROR;
}
return OPENSFS_SUCC;
}
//this function is much to slow on a genome scale should be improved
fMap getMap(char *fname,std::map<char*,int,ltstr> *revMap){
const char *delims = "\t\n ";
FILE *fp=getFILE(fname,"r");
char buf[LENS];
int nsites=0;
fMap ret;
std::map<char *,int>::iterator rit;
while(fgets(buf,LENS,fp)){
char *chr = strtok(buf,delims);
strtok(NULL,delims);//rsnumber
strtok(NULL,delims);//centimorgan
char *tok = strtok(NULL,delims);
if(tok==NULL){
fprintf(stderr,"Problem with fileformat in .bim file\n");
exit(0);
}
int pos = atoi(tok)-1;//genomic position in bp
mm value;
value.major = refToInt[strtok(NULL,delims)[0]];
value.minor = refToInt[strtok(NULL,delims)[0]];
//check for N if this exists;
if(value.major==4||value.minor==4){
fprintf(stderr,"N extists in major minor defintion\n");
break;
}
// fprintf(stderr,"chr=%s pos=%d major=%d minor=%d\n",chr,pos,mymm.major,mymm.minor);
rit=revMap->find(chr);
if(rit==revMap->end()){
fprintf(stderr,"Problem finding chromosome: %s in lookuptable\n",chr);
exit(0);
}
mm key;
key.major = rit->second;
key.minor = pos;
fMap::iterator it = ret.find(key);
if(it!=ret.end()){
fprintf(stderr,"duplicate entry in filterlist:%s : will exit offending position below\n",fname);
fprintf(stderr,"chr=%s pos=%d major=%d minor=%d\n",chr,pos,value.major,value.minor);
exit(0);
}else
ret.insert(fMap::value_type(key, value));
}
fclose(fp);
return ret;
}
void readSFS(const char*fname,int hint,double *ret){
fprintf(stderr,"reading: %s\n",fname);
FILE *fp = getFILE(fname,"r");
char buf[fsize(fname)+1];
if(fsize(fname)!=fread(buf,sizeof(char),fsize(fname),fp)){
fprintf(stderr,"Problems reading file: %s\n will exit\n",fname);
exit(0);
}
buf[fsize(fname)]='\0';
std::vector<double> res;
char *tok=NULL;
tok = strtok(buf,"\t\n ");
if(!tok){
fprintf(stderr,"File:%s looks empty\n",fname);
exit(0);
}
res.push_back(atof(tok));
while((tok=strtok(NULL,"\t\n "))) {
//fprintf(stderr,"%s\n",tok);
res.push_back(atof(tok));
}
// fprintf(stderr,"size of prior=%lu\n",res.size());
if(hint!=res.size()){
fprintf(stderr,"\t-> Problem with size of dimension of prior %d vs %lu\n",hint,res.size());
for(size_t i=0;0&&i<res.size();i++)
fprintf(stderr,"%zu=%f\n",i,res[i]);
exit(0);
}
for(size_t i=0;i<res.size();i++){
ret[i] = exp(res[i]);
// fprintf(stderr,"i=%d %f\n",i,ret[i]);
}
fclose(fp);
}
void filter::getOptions(argStruct *arguments){
fname=angsd::getArg("-filter",fname,arguments);
if(fname!=NULL)
doFilter = findType(fname);
//1=bim 2=keep
doMajorMinor = angsd::getArg("-doMajorMinor",doMajorMinor,arguments);
if(doMajorMinor==3 && doFilter!=1){
fprintf(stderr,"Must supply -filter with .bim file if -doMajorMinor 3\n");
exit(0);
}
if(doFilter==1){
fm = getMap(fname,revMap);
fprintf(stderr,"\t-> number of sites in filter: %lu\n",fm.size());
}else if(doFilter==2){
fp = getFILE(fname,"r");
// readSites();
fprintf(stderr,"Filtering with .keep is still beta\n");
}
minInd = angsd::getArg("-minInd",minInd,arguments);
}
int main (int argc, char *argv[]) {
if (argc==1) {
info();
return 0;
}
/// DECLARE AND INITIALIZE VARIABLES
char *sfsfile1=NULL; // posterior probabilities
char *sfsfile2=NULL;
char *fstfile=NULL; // first guess of fst
char *priorfile1=NULL; // priors (needed for weighting function only)
char *priorfile2=NULL;
char *priorfile12=NULL; // joint prior, it is 2D-SFS
FILE *outpost;
char *outfile=NULL;
char *foutpost=NULL;
int argPos = 1, increment = 0, nind = 0, nind1 = 0, nind2 = 0, nsites = 0, verbose = 0, nsums = 1, block_size = 10000, K=0, isfold=0, firstbase=0;
/// READ AND ASSIGN INPUT PARAMETERS
while (argPos<argc) {
increment = 0;
if(strcmp(argv[argPos],"-postfiles")==0) {
sfsfile1 = argv[argPos+1];
sfsfile2 = argv[argPos+2];
increment = 1;
}
else if(strcmp(argv[argPos],"-fstfile")==0) {
fstfile = argv[argPos+1];
}
else if(strcmp(argv[argPos],"-priorfile")==0) {
priorfile12 = argv[argPos+1];
}
else if(strcmp(argv[argPos],"-priorfiles")==0) {
priorfile1 = argv[argPos+1];
priorfile2 = argv[argPos+2];
increment = 1;
}
else if(strcmp(argv[argPos],"-outfile")==0) outfile = argv[argPos+1];
else if(strcmp(argv[argPos],"-nind")==0) {
nind1 = atoi(argv[argPos+1]);
nind2 = atoi(argv[argPos+2]);
nind = nind1 + nind2;
increment = 1;
}
else if(strcmp(argv[argPos],"-nsites")==0) nsites = atoi(argv[argPos+1]);
else if(strcmp(argv[argPos],"-K")==0) K = atof(argv[argPos+1]);
else if(strcmp(argv[argPos],"-verbose")==0) verbose = atoi(argv[argPos+1]);
else if(strcmp(argv[argPos],"-block_size")==0) block_size = atoi(argv[argPos+1]);
else if(strcmp(argv[argPos],"-nsums")==0) nsums = atoi(argv[argPos+1]);
else if(strcmp(argv[argPos],"-firstbase")==0) firstbase = atoi(argv[argPos+1]);
else if(strcmp(argv[argPos],"-isfold")==0) isfold = atoi(argv[argPos+1]);
else {
printf("\tUnknown arguments: %s\n",argv[argPos]);
info();
return 0; // terminate
}
argPos = argPos + 2 + increment;
}
/// CHECK INPUT
if((sfsfile1 == NULL) & (sfsfile2 == NULL) ) {
fprintf(stderr,"\nMust supply -postfiles.\n");
info();
return 0;
}
if(outfile == NULL) {
fprintf(stderr,"\nMust supply -outfile.\n");
info();
return 0;
}
if((priorfile1 == NULL) & (fstfile==NULL) & (K==0)) {
fprintf(stderr,"\nPerhaps you forgot to supply -priofiles and -fstfile when using an automatic setting of lambda?\n");
}
if((priorfile1 != NULL) & (priorfile12!=NULL)) {
fprintf(stderr,"\nYou should give either -priorfiles or -priorfile, otherwise I don't know if you want to use a 2D-SFS or the corrected product of marginal spectra as prior\n");
info();
return 0;
}
if((fstfile != NULL) & (priorfile12!=NULL)) {
fprintf(stderr,"\nIf you give -fstfile I assume you want to use the correction for marginal spectra. So why are you giving -priorfile too? You should only give -priorfiles eventually if K=0.\n");
info();
return 0;
}
if((isfold) & (priorfile12!=NULL)) {
fprintf(stderr,"\nSorry. Handling the folded 2D-SFS has not been implemented yet. Please contribute or push a request. Currently -ifold 1 and -priorfile12 !NULL are not compatible.\n");
info();
return 0;
}
/// OUTPUT
foutpost = append(outfile, "");
fprintf(stderr,"\t->Dumping file: %s\n", foutpost);
outpost = getFILE(foutpost, "w");
// print input arguments // UPDATE THIS !!!
fprintf(stderr,"\t->Using some of these args: -nind %d -nind1 %d -nind2 %d -nsites %d -postfiles %s %s -priorfiles %s %s -priorfile %s -fstfile %s -outfile %s -verbose %d -nsums %d -offset %d -K %d\n", nind, nind1, nind2, nsites, sfsfile1, sfsfile2, priorfile1, priorfile2, priorfile12, fstfile, foutpost, verbose, nsums, firstbase, K);
// READ PRIORS (if provided)
// marginal spectra
array<double> prior1;
array<double> prior2;
if (priorfile1 != NULL) {
if (verbose==1) fprintf(stderr, "\nAdding priors...");
prior1 = readArray(priorfile1, nind1, isfold);
prior2 = readArray(priorfile2, nind2, isfold);
}
// 2D-SFS
matrix<double> prior12;
if ((priorfile12==NULL)==0) {
if (verbose==1) fprintf(stderr, "\nAdding 2D prior...");
prior12 = readPrior12(priorfile12, nind1*2+1, nind2*2+1);
if (verbose==2) {
fprintf(stderr, "\nPrior 2d:\n");
writematrix(prior12, stderr);
}
//// the difference with this prior is that I don't add the prior, but I add the prior directly at computeFST step
}
/// GET POSITIONS OF BLOCKS
if (block_size>(nsites-firstbase)) block_size=(nsites-firstbase);
array<int> start; array<int> end;
start=getStart(nsites, firstbase, block_size);
end=getEnd(nsites, firstbase, block_size);
/// ITERATE OVER EACH BLOCK
int nwin = start.x;
for (int n=0; n<nwin; n++) {
fprintf(stderr, "Block %d out of %d from %d to %d\n", n, (nwin-1), start.data[n], end.data[n]);
// READ POSTERIOR PROBABILITIES FILES
matrix<double> post1;
matrix<double> post2;
post1 = readFileSub(sfsfile1, nind1, start.data[n], end.data[n], isfold);
post2 = readFileSub(sfsfile2, nind2, start.data[n], end.data[n], isfold);
if (priorfile12!=NULL) {
// if from -realSFS 1 they are -log
normSFS(post1, 1); // 2nd argument is islog
normSFS(post2, 1);
}
// IF NOT FST FILE PROVIDED
if ((fstfile == NULL)) {
if (verbose==1) fprintf(stderr,"Computing FST and no first guess provided.\n");
if (priorfile12==NULL) {
if (isfold) {
computeVarReyFold(post1, post2, verbose, outpost, nsums);
} else {
computeVarRey(post1, post2, verbose, outpost, nsums);
}
} else {
if (verbose==1) fprintf(stderr,"Using 2D-SFS as prior. You didn't run sfstools, right??? Use only -realSFS 1.\n");
computeVarRey12New(post1, post2, verbose, outpost, nsums, prior12);
}
} else {
// IF FST FILE IS INDEED PROVIDED
if (verbose==1) fprintf(stderr,"Computing FST and first guess provided.\n");
array<double> firstfst;
firstfst=readFSTsub(fstfile, nsites, start.data[n], end.data[n]);
array <double> sublam;
sublam = getLambdas(firstfst, prior1, prior2, K, verbose, isfold);
if (verbose==1) fprintf(stderr,"Computed lambdas.\n");
if (isfold) {
computeVarRey2Fold(post1, post2, verbose, outpost, nsums, sublam);
} else {
computeVarRey2(post1, post2, verbose, outpost, nsums, sublam);
}
delete [] firstfst.data;
delete [] sublam.data;
}
cleanup(post1);
cleanup(post2);
} // end blocks iterations
delete [] start.data;
delete [] end.data;
free(foutpost);
return 0;
} // end main
