impute_bulk_met: Impute/predict bulk methylation states
In BPRMeth: Model higher-order methylation profiles

Description Usage Arguments Value Author(s) See Also Examples

Make predictions of missing methylation states, i.e. perfrom imputation using BPRmeth This requires keepin a subset of data as a held out test set during BPRMeth inference or providing a different file that contains chromosome and CpG locations.

1	impute_bulk_met(obj, anno, test_data = NULL, return_test = FALSE)

`obj`	Output of BPRMeth inference object.
`anno`	A `GRanges` object with annotation data, whose format should be similar to `read_anno`.
`test_data`	Test data to evaluate performance.
`return_test`	Whether or not to return a list with the predictions.

A list containing two vectors, the true methylation state and the predicted/imputed methylation states.

C.A.Kapourani C.A.Kapourani@ed.ac.uk

partition_bulk_dataset

# Extract synthetic data
dt <- encode_met

# Partition to train and test set
dt <- partition_bulk_dataset(dt)

# Create basis object
basis_obj <- create_rbf_object(M = 3)

# Run BPRMeth
fit <- infer_profiles_mle(X = dt$met, model = "binomial",
   basis = basis_obj, is_parallel = FALSE, opt_itnmax = 10)

# Perform imputation
imputation_obj <- impute_bulk_met(obj = fit, anno = dt$anno,
                                  test_data = dt$met_test)