v2.22.0

MIToS v2.22.0

Diff since v2.21.0

This versions introduces several breaking changes to improve the usability of the
Information module. The main changes are:

• [Breaking change] The Information module deprecates the Counts type in favor of
  the new Frequencies type. The new type as the same signature and behavior as the old one.

• [Breaking change] The count function on sequences has been deprecated in favor of the
  frequencies function, which has the same signature and behavior as the old one.

• [Breaking change] The count! function is deprecated in favor of frequencies!.
  The new function use keyword arguments to define the weights and pseudocounts. As an
  example of migration, count!(table, weights, pseudocounts, seqs...) should be replaced
  by frequencies!(table, seqs..., weights=weights, pseudocounts=pseudocounts).

• [Breaking change] The probabilities! method using positional arguments for the
  weights, pseudocounts and pseudofrequencies is deprecated in favor the one that uses
  keyword arguments. As an example of migration,
  probabilities!(table, weights, pseudocounts, pseudofrequencies, seqs...)
  should be replaced by
  probabilities!(table, seqs..., weights=weights, pseudocounts=pseudocounts, pseudofrequencies=pseudofrequencies).

• [Breaking change] The Information has deprecated the entropy method on
  Frequencies and Probabilities in favor of the shannon_entropy function. The
  definition of the base is now done using the base keyword argument. As an example of
  migration, entropy(p, 2) should be replaced by shannon_entropy(p, base=2).

• [Breaking change] The marginal_entropy methods based on positional arguments are
  deprecated in favor of a method relying on the margin and base keyword arguments.
  As an example of migration, marginal_entropy(p, 2, 2.0) should be replaced by
  marginal_entropy(p, margin=2, base=2.0).

• [Breaking change] The mutual_information method based on positional arguments is
  deprecated in favor of a method relying on the base keyword argument. As an example of
  migration, mutual_information(p, 2) should be replaced by mutual_information(p, base=2).

• [Breaking change] The mapcolpairfreq! and mapseqpairfreq! functions now uses the
  boolean usediagonal keyword argument to indicate if the function should be applied to
  the diagonal elements of the matrix (the default is true). Before, this was done passing
  Val{true} or Val{false} as the last positional argument.

• The mapcolfreq!, mapseqfreq!, mapcolpairfreq!, and mapseqpairfreq! methods using
  keyword arguments, now pass the extra keyword arguments to the mapped function.

• The Information module now exports the mapfreq function that offers a more high-level
  interface to the mapcolfreq!, mapseqfreq!, mapcolpairfreq!, and mapseqpairfreq!
  functions. This function allows the user to map a function to the residue frequencies or
  probabilities of the columns or sequences of an MSA. When rank = 2, the function is
  applied to pairs of sequences or columns.

• The Information module now exports methods of the shannon_entropy, kullback_leibler,
  mutual_information, and normalized_mutual_information functions that take an
  AbstractArray{Residue} as input, e.g. an MSA. Those methods use the mapfreq function
  under the hood to ease the calculation of the information measures on MSAs.

• The frequencies!, frequencies, probabilities!, and probabilities functions now
  accept arrays of Residues of any dimension. Therefore, there is no need to use the
  vec function to convert the arrays to vectors.

• The MSA module now exports the WeightType union type to represent weights.