void ErrorRateEstimator::Estimate()
{
// First we try a genotyping a model with genotyping error
//
printf("Fitting Maximum Likelihood Estimate for Error Rate...\n");
alleleError = true;
FuzzyInheritanceTree::perGenotypeError = 0.0;
EstimateModel();
double perAlleleEstimate = min;
double perAlleleLk = fmin;
alleleError = false;
FuzzyInheritanceTree::perAlleleError = 0.0;
EstimateModel();
if (perAlleleLk < fmin)
{
printf("The per allele error model fits best.\n"
" Estimated error rate = %.4f per allele\n\n",
perAlleleEstimate);
FuzzyInheritanceTree::perGenotypeError = 0.0;
FuzzyInheritanceTree::perAlleleError = perAlleleEstimate;
}
else
{
printf("The per genotype error model fits best.\n"
" Estimated error rate = %.4f per genotype\n\n",
min);
FuzzyInheritanceTree::perGenotypeError = min;
FuzzyInheritanceTree::perAlleleError = 0.0;
}
}
int main(int argc, char *argv[])
{
char *datafn, *s;
int nSeg;
void Initialise(void);
void LoadFile(char *fn);
void EstimateModel(void);
void SaveModel(char *outfn);
if(InitShell(argc,argv,hinit_version,hinit_vc_id)<SUCCESS)
HError(2100,"HInit: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(2100,"HInit: InitParm failed");
InitTrain(); InitUtil();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
CreateHMMSet(&hset,&gstack,FALSE);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(2119,"HInit: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'e':
epsilon = GetChkedFlt(0.0,1.0,s); break;
case 'i':
maxIter = GetChkedInt(0,100,s); break;
case 'l':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Segment label expected");
segLab = GetStrArg();
break;
case 'm':
minSeg = GetChkedInt(1,1000,s); break;
case 'n':
newModel = FALSE; break;
case 'o':
outfn = GetStrArg();
break;
case 'u':
SetuFlags(); break;
case 'v':
minVar = GetChkedFlt(0.0,10.0,s); break;
case 'w':
mixWeightFloor = MINMIX * GetChkedFlt(0.0,10000.0,s);
break;
case 'B':
saveBinary = TRUE;
break;
case 'F':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Data File format expected");
if((dff = Str2Format(GetStrArg())) == ALIEN)
HError(-2189,"HInit: Warning ALIEN Data file format set");
break;
case 'G':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Label File format expected");
if((lff = Str2Format(GetStrArg())) == ALIEN)
HError(-2189,"HInit: Warning ALIEN Label file format set");
break;
case 'H':
if (NextArg() != STRINGARG)
HError(2119,"HInit: HMM macro file name expected");
AddMMF(&hset,GetStrArg());
break;
case 'I':
if (NextArg() != STRINGARG)
HError(2119,"HInit: MLF file name expected");
LoadMasterFile(GetStrArg());
break;
case 'L':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Label file directory expected");
labDir = GetStrArg(); break;
case 'M':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Output macro file directory expected");
outDir = GetStrArg();
break;
case 'T':
if (NextArg() != INTARG)
HError(2119,"HInit: Trace value expected");
trace = GetChkedInt(0,01777,s);
break;
case 'X':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Label file extension expected");
labExt = GetStrArg(); break;
default:
HError(2119,"HInit: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(2119,"HInit: source HMM file name expected");
hmmfn = GetStrArg();
Initialise();
do {
if (NextArg()!=STRINGARG)
HError(2119,"HInit: training data file name expected");
datafn = GetStrArg();
LoadFile(datafn);
} while (NumArgs()>0);
nSeg = NumSegs(segStore);
if (nSeg < minSeg)
HError(2121,"HInit: Too Few Observation Sequences [%d]",nSeg);
if (trace&T_TOP) {
printf("%d Observation Sequences Loaded\n",nSeg);
fflush(stdout);
}
EstimateModel();
SaveModel(outfn);
if (trace&T_TOP)
printf("Output written to directory %s\n",
outDir==NULL?"current":outDir);
Exit(0);
return (0); /* never reached -- make compiler happy */
}
int mainHInit(int argc, char *argv[])
{
static int ft=1;
char *datafn, *s;
int nSeg;
zwangModify();
if(InitShell(argc,argv,hinit_version,hinit_vc_id)<SUCCESS)
HError(2100,"HInit: InitShell failed");
if(ft)
{
if(isMemInit==0)
{
InitMem();
isMemInit=1;
}
InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(2100,"HInit: InitParm failed");
InitTrain(); InitUtil();
ft=0;
/* Stacks for global structures requiring memory allocation */
CreateHeap(&segmentStack,"SegStore", MSTAK, 1, 0.0, 100000, LONG_MAX);
CreateHeap(&sequenceStack,"SeqStore", MSTAK, 1, 0.0, 1000, 1000);
CreateHeap(&clustSetStack,"ClustSetStore", MSTAK, 1, 0.0, 1000, 1000);
CreateHeap(&transStack,"TransStore", MSTAK, 1, 0.0, 1000, 1000);
CreateHeap(&traceBackStack,"TraceBackStore", MSTAK, 1, 0.0, 1000, 1000);
CreateHeap(&bufferStack,"BufferStore", MSTAK, 1, 0.0, 1000, 1000);
}
zwangHMemGetConf();
zwangHWaveGetConf();
zwangHLabelGetConf();
zwangHMathGetConf();
zwangHSigPGetConf();
zwangHAudioGetConf();
zwangHVQGetConf();
zwangHModelGetConf();
zwangHParmGetConf();
zwangHTrainGetConf();
zwangHUtilGetConf();
SetConfParms();
CreateHMMSet(&hset,&gstack,FALSE);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(2119,"HInit: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'e':
epsilon = GetChkedFlt(0.0,1.0,s); break;
case 'i':
maxIter = GetChkedInt(0,100,s); break;
case 'l':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Segment label expected");
segLab = GetStrArg();
break;
case 'm':
minSeg = GetChkedInt(1,1000,s); break;
case 'n':
newModel = FALSE; break;
case 'o':
outfn = GetStrArg();
break;
case 'u':
SetuFlags(); break;
case 'v':
minVar = GetChkedFlt(0.0,10.0,s); break;
case 'w':
mixWeightFloor = MINMIX * GetChkedFlt(0.0,10000.0,s);
break;
case 'B':
saveBinary = TRUE;
break;
case 'F':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Data File format expected");
if((dff = Str2Format(GetStrArg())) == ALIEN)
HError(-2189,"HInit: Warning ALIEN Data file format set");
break;
case 'G':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Label File format expected");
if((lff = Str2Format(GetStrArg())) == ALIEN)
HError(-2189,"HInit: Warning ALIEN Label file format set");
break;
case 'H':
if (NextArg() != STRINGARG)
HError(2119,"HInit: HMM macro file name expected");
AddMMF(&hset,GetStrArg());
break;
case 'I':
if (NextArg() != STRINGARG)
HError(2119,"HInit: MLF file name expected");
LoadMasterFile(GetStrArg());
break;
case 'L':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Label file directory expected");
labDir = GetStrArg(); break;
case 'M':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Output macro file directory expected");
outDir = GetStrArg();
break;
case 'T':
if (NextArg() != INTARG)
HError(2119,"HInit: Trace value expected");
trace = GetChkedInt(0,01777,s);
break;
case 'X':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Label file extension expected");
labExt = GetStrArg(); break;
default:
HError(2119,"HInit: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(2119,"HInit: source HMM file name expected");
hmmfn = GetStrArg();
Initialise();
do {
if (NextArg()!=STRINGARG)
HError(2119,"HInit: training data file name expected");
datafn = GetStrArg();
LoadFile(datafn);
} while (NumArgs()>0);
nSeg = NumSegs(segStore);
if (nSeg < minSeg)
HError(2121,"HInit: Too Few Observation Sequences [%d]",nSeg);
EstimateModel();
SaveModel(outfn);
Dispose(hset.hmem,hset.mtab);
ResetHeap(&gstack);
ResetHeap(&segmentStack);
ResetHeap(&sequenceStack);
ResetHeap(&clustSetStack);
ResetHeap(&transStack);
ResetHeap(&traceBackStack);
ResetHeap(&bufferStack);
zwangInitParmClear();
zwangInitShellClear();
return (0); /* never reached -- make compiler happy */
}
