This function determines the hyperparameters for the Bayesian priors based on the input data.

```
auto_config_run(
x,
K = c(1:3),
NB_size_atac = 150,
NB_size_rna = 150,
a_sd = 0.1,
b_sd = 1,
prior_cn = c(0.1, 0.6, 0.1, 0.1, 0.1),
hidden_dim = 5,
init_importance = 0.6,
NB_size_priors = c(15, 1000),
CUDA = FALSE,
normal_cells = FALSE
)
```

## Arguments

- x
(required) CONGAS+ object.

- K
(required) Number of clusters that will be tested during the inference.

- NB_size_atac
Float (optional). Default 150. Value used to inizialize the size hyperparametr of the Negative Binomial for ATAC in case the likelihood for ATAC is set to NB

- NB_size_rna
Float (optional). Default 150. Value used to inizialize the size hyperparametr of the Negative Binomial for RNA in case the likelihood for RNA is set to NB

- a_sd
Float (optional). Default 0.1. Lower bound of the Uniform prior for the Gussian standard deviation. Used when one of RNA or ATAC likelihoods are gaussian.

- b_sd
Float (optional). Default 1. Upper bound of the Uniform prior for the Gussian standard deviation. Used when one of RNA or ATAC likelihoods are gaussian.

- prior_cn
(Optional) Default c(0.1, 0.6, 0.1, 0.1, 0.1). Prior for the copy number state of every segment.

- hidden_dim
(Optional) defualt 5. Number of discrete copy number states to model. By default this is from 1 to 5.

- init_importance
(Optional) Default 0.6. Value used to initialize the distribution over possible copy number states for every cluster and every segment. `init_importance`

is used to initialize the value corresponding to the copy number state inferred from bulk and `1-init_importance / (hidden_dim - 1)`

is used to initialize the other ploidy states.
This distribution is initialized based on the value of init importance, and then its prior is defined in `prior_cn`

.

- NB_size_priors
(optional) Default c(15, 1000). Lower and upper bound of the Uniform prior on the Negative binomial size hyperparameter.

- CUDA
Defualt FALSE. Flag indicating whether to use GPU computation.

- normal_cells
Default to FALSE. Flag that can be used to inject prior knowledge about the presence of normal cells in the sample. In case this is set to TRUE, the copy number
distribution for one of the clusters will be initialized with values skewed towards the diploid state in every segment.