rasqual:
  RASQUAL (Robust Allele Specific Quantitation and quality controL) maps QTLs
  using population and AS signals for any sequenced based cellular tratis.

Usage:
  VCF as stdin | ./rasqual [OPTIONS] -y <counts.bin> -k <offsets.bin> -n <sample_size> -j <feature_id> \
                                     -l <#tested_SNPs> -m <#feature_SNPs> -s <feature_start(s)> -e <feature_end(s)>
									 
Output:
  Tested SNPs in stdout and posterior genotype in VCF file

Mandatory arguments:

  -y/--feature-counts            <string>  File name of total feature counts in binary double
  -k/--sample-offsets            <string>  File name of sample specific offset terms in binary double
  -s/--feature-starts            <string>  Feature start position(s) in comma separated value (e.g., 100,300,500)
  -e/--feature-ends              <string>  Feature end position(s) in comma separated value (e.g., 150,350,450)
  -j/--feature-id                <int>     Row number of target feature in original count table
  -l/--number-of-testing-snps    <int>     Number of testing SNPs in VCF (candidates of a rSNP)
  -m/--number-of-fsnps           <int>     Number of feature SNPs (fSNPs) in VCF
  -n/--sample-size               <int>     Sample size
  
Output options:
  
  -t/--lead-snp                            Output only the lead QTL SNP
  -i/--imprint                             Output imprinted (or randomly allelicly inactivated) feature result
  -g/--posterior-genotype        <string>  Output posterior genotypes as VCF format
  -f/--feature-name              <string>  Feature name printed on the result (default: feature ID number)

Test SNP (rSNP) filter options:

  -c/--cis-midpoint              <int>     Midpoint of cis regulatory window, e.g., TSS (default: midpoint of feature start and end)
  -w/--cis-window-size           <int>     Cis regulatory window size centered at the midpoint (default: 54e7)
  -a/--minor-allele-frequency    <double>  Minor allele frequency cutoff (default: 0.05)
  -h/--hardy-weinberg-pvalue     <double>  Hardy-Weinberg equiliblium P-value cutoff (default: 1.0e-8)
  -q/--imputation-quality        <double>  Genome imputation quality score cutoff, e.g., R^2 (MaCH/Beagle) or I^2 (IMPUTE) (default: 0.7)
  --rsnp                         <string>  Test only specific SNP (variant ID given in the VCF file)

fSNP filter options:

  --minor-allele-frequency-fsnp  <double>  Minor allele frequency cutoff for fSNPs (default: 0.0)
  --hardy-weinberg-pvalue-fsnp   <double>  Hardy-Weinberg equiliblium P-value cutoff for fSNPs (default: 0.0)
  --imputation-quality-fsnp      <double>  Genome imputation quality score cutoff for fSNPs (default: 0.0)
  --min-coverage-depth           <double>  Minimum depth to filter out low coverage fSNPs (default: 0.05)
  --effective-feature-length     <int>     Effective length of feature to calculate average depth (default: length of union features)

Model fitting options:

  -x/--covariates                <string>  File name of covariates in bianary double
  -p/--number-of-covariates      <int>     Number of covariates used for adjustment
  -r/--random-permutation                  Randomly permute samples
  -k2/--conditional-snps         <string>  Conditional SNP(s) ID with effect size(s) connected by colon (:) (e.g., rs1:0.12,rs2:0.34) 
  --population-only                        Only total feature counts are modelled
  --as-only                                Only AS counts are modelled
  --null                                   Fitting null model only
  --force                                  Fitting model for the large numbers of fSNPs and rSNPs (#fSNPs x #rSNPs < 30,000 w/o this option)
  --fix-delta                              Fix sequencing/mapping error parameter (Delta=0.01)
  --fix-phi                                Fix reference allele mapping parameter (Phi=0.5)
  --fix-theta                              Fix overdispersion parameter (Theta=10000)
  --number-of-significant-loci   <int>     Average number of SNPs tested by joint model (default: number of tested SNPs)
  --n-threads                    <int>     The number of threads used for mapping QTLs in parallel

Genotype uncertainty options:
  
  -z/--convert-imputation-score            Convert genome imputation quality score (R^2 or I^2) into allelic probability (AP)
  --nominal-allelic-probability            Use nomial AP (default: AP is truncated in [0.001, 0.999] to allow uncertainty)
  --genotype-dosage                        Use genotype dosage as AP
  --population-allele-frequency            Use population allele frequency as AP
  --fix-genotype                           Use fix genotype (in GT field)
  --no-posterior-update                    Not updating posterior probability
  
Other options:

  -v/-V   Print verbose output (for debugging). 

Hyperparameter options (modification unrecomended):

  -SIGMA  <double>    Variance parameter for Normal gamma distribution on Lambda (default: 10000)
  -BETA   <double>    Mean parameter for Normal-gamma distribution on Lambda (default: sample mean)
  -ABPI   <double>    Shape parameters (identical) for Beta distribution on Pi (default:10)
  -ABPHI  <double>    Shape parameters (identical) for Beta distribution on Phi (default:10)
  -ADELTA <double>    1st shape parameter for Beta distribution on Delta (default: 1.01)
  -BDELTA <double>    2nd shape parameter for Beta distribution on Delta (default: 1.99) 
  -KAPPA  <double>    Shape parameter for Gamma distribution on Theta (default: 2.02)
  -OMEGA  <double>    Rate parameter for Gamma distribution on Theta (default: 0.2)

#
# not implemented
#  -hetType
#  -trans
#