find_prior_weights: Find prior weights
In andrewGhazi/malacoda: Bayesian Analysis of High-Throughput Genomic Assays

For a given variant and annotation set (scaled) and distance matrix, get the weights of all other variants

find_prior_weights(
  given_id,
  scaled_annotations,
  dist_mat,
  min_dist_kernel,
  kernel_fold_change = 1.3,
  min_num_neighbors = 100,
  verbose = TRUE
)

`given_id`	a variant_id to get weights for
`scaled_annotations`	a tall annotation data frame where the annotations have been set to the same scale
`dist_mat`	a distance matrix of euclidean distances between variants in scaled annotation space
`min_dist_kernel`	an initialization of the distance kernel at some tiny value
`kernel_fold_change`	the multiplier by which to iteratively increase the distance kernel
`min_num_neighbors`	the minimum number of neighbors which must make a meaningful contribution to the weights before stopping
`verbose`	logical indicating whether to print messages

The "meaningful contribution" is defined in this way: The variants are sorted by weight. The min_num_neighbors-th (e.g. 100th) variant will be weighted to at least 1 prevents some small number of extremely close neighbors from dominating the prior estimation later on.

andrewGhazi/malacoda documentation built on Aug. 2, 2020, 12:54 a.m.