updog_update_all: This is the main optimization function for updog.
In dcgerard/updogAlpha: Using Parental Data for Offspring Genotyping

Description Usage Arguments Value Author(s)

This is the main optimization function for updog.

updog_update_all(ocounts, osize, ploidy, p1counts = NULL, p1size = NULL,
  p2counts = NULL, p2size = NULL, print_val = TRUE, tol = 10^-4,
  maxiter = 500, commit_num = 4, min_disp = 0, update_outmean = FALSE,
  update_outdisp = FALSE, update_outprop = TRUE, update_bias_val = TRUE,
  update_seq_error = TRUE, update_od_param = TRUE, update_pgeno = TRUE,
  p1geno = round(ploidy/2), p2geno = round(ploidy/2), seq_error = 0.01,
  od_param = 0.01, bias_val = 1, out_prop = 0.001, out_mean = 0.5,
  out_disp = 1/3, non_mono_max = 2, bound_bias = FALSE,
  seq_error_mean = -4.7, seq_error_sd = 1, bias_val_mean = 0,
  bias_val_sd = 0.7, allele_freq = 0.5, model = c("f1", "s1", "hw",
  "uniform"), verbose = FALSE)

`ocounts`	A vector of non-negative integers. The ith element is the number of reads of the reference allele in the ith child.
`osize`	A vector of positive integers. The ith element is the total number of reads for the ith child.
`ploidy`	A positive integer. The number of copies of the genome in the species. This is the assumed to be the same for all individuals.
`p1counts`	A vector of non-negative integers. The ith element is the number of reads of the reference allele in the ith sample of parent 1. If `NULL` then the prior probabilities on parent 1's genotype will default to uniform.
`p1size`	A vector of positive integers. The ith element is the total number of reads in the ith sample of parent 1. If `NULL` then the prior probabilities on parent 1's genotype will default to uniform.
`p2counts`	A vector of non-negative integers. The ith element is the number of reads of the reference allele in the ith sample of parent 2. If `NULL` then the prior probabilities on parent 2's genotype will default to uniform.
`p2size`	A vector of positive integers. The ith element is the total number of reads in the ith sample of parent 2. If `NULL` then the prior probabilities on parent 2's genotype will default to uniform.
`print_val`	A logical. Should we print the updates (`TRUE`) or not (`FALSE`)?
`tol`	The stopping criterion
`maxiter`	The maximum number of iterations
`commit_num`	The number of consecutive iterations where the parental genotypes do not change before we commit to those parental genotypes.
`min_disp`	The minimum value for the over-dispersion parameter of the outlier distribution. If this gets too small, then we can be overly confident in points being outliers.
`update_outmean`	A logical. Should we update `out_mean` (`TRUE`) or not (`FALSE`)?
`update_outdisp`	A logical. Should we update `out_mean` (`TRUE`) or not (`FALSE`)?
`update_outprop`	A logical. Should we update `out_prop` (`TRUE`) or not (`FALSE`)?
`update_bias_val`	A logical. Should we update `bias_val` (`TRUE`) or not (`FALSE`)?
`update_seq_error`	A logical. Should we update `seq_error` (`TRUE`) or not (`FALSE`)?
`update_od_param`	A logical. Should we update `od_param` (`TRUE`) or not (`FALSE`)?
`update_pgeno`	A logical. Should we update `p1geno` and `p1geno` (`TRUE`) or not (`FALSE`)?
`p1geno`	The initial value of the first parental genotype.
`p2geno`	The initial value of the second parental genotype.
`seq_error`	The initial value of the sequencing error rate.
`od_param`	The initial value of the overdispersion parameter.
`bias_val`	The initial value of the bias parameter.
`out_prop`	The initial value of the proportion of points that are outliers.
`out_mean`	The initial value of the mean of the outlier distribution.
`out_disp`	The initial value of the over-dispersion parameter of the outlier distribution.
`non_mono_max`	The maximum number of iterations to allow non-monotonicity of likelihood.
`bound_bias`	A logical. Should we bound `bias_val` by a somewhat arbitrary value (`TRUE`) or not (`FALSE`)?
`seq_error_mean`	The mean of the logit-normal prior on the sequencing error rate, which corresponds to `parvec[2]`. Set `seq_error_sd = Inf` to have no penalty on the sequencing error rate. The default is -4.7, which roughly corresponds to a mean sequencing error value of 0.009. If you want to constain `seq_error` to be zero, you need to set `update_seq_error = FALSE`, `seq_error = 0`, and`seq_error_mean = -Inf`.
`seq_error_sd`	The standard deviation of the logit-normal prior on the sequencing error rate, which corresponds to `parvec[2]`. The default is 1, which at three standard deviations is about a sequencing error rate of 0.15. Set `seq_error_sd = Inf` to have no penalty on the sequencing error rate.
`bias_val_mean`	The prior mean on the log of `bias_val` (corresponding to `parvec[1]`). Set `bias_val_sd = Inf` to have no penalty on the bias parameter.
`bias_val_sd`	The prior standard deviation on the log of `bias_val` (corresponding to `parvec[1]`). Set `bias_val_sd = Inf` to have no penalty on the bias parameter.
`allele_freq`	The allele frequency if `model = "hw"`.
`model`	The model for the genotype distribution. Do we assume an F1 population (`"f1"`), an S1 population (`"s1"`), Hardy-Weinberg equilibrium (`"hw"`), or a uniform distribution (`"uniform"`).
`verbose`	A logical. Should we output a lot more (`TRUE`) or not (`FALSE`)?

A list of the following elements

bias_val: The estimated bias parameter.
seq_error: The estimated sequencing error rate.
od_param: The estimated overdispersion parameter.
p1geno: The estimated genotype of one parent.
p1geno: The estimated genotype of the other parent.
out_prop: The estimated proportion of points that are outliers.
out_mean: The estimated mean of the outlier distribution.
out_disp: The estimated overdispersion parameter of the outlier distribution.
prob_out: A vector. Each element of which is the posterior probability that a point is an outlier.
allele_freq: The estimated allele-frequency of the reference allele.
p1_prob_out: The posterior probability that parent 1 is an outlier.
p2_prob_out: The posterior probability that parent 2 is an outlier.
num_iter: The number of iterations the optimization program was run.
convergence: 1 is we reached maxiter and 0 otherwise.
llike: The final log-likelihood of the estimates.
hessian: The Fisher information under the parameterization (s, ell, r), where s = log(bias_val) = log(d), ell = logit(seq_error) = logit(eps), and r = - logit(od_param) = - logit(tau).