layer_deNovo: Custom 2-D convolution wrapper with to learn IGMs or PWMs.
In mPloenzke/learnMotifs: Learn Nucleotide Motifs via Convolutions

Description Usage Arguments Value Author(s)

Custom 2-D convolution wrapper with regularization and constraints to learn de novo motifs in the form of IGMs or PWMs.

layer_deNovo(object, filters, filter_len, lambda_pos = 0,
  lambda_filter = 0, lambda_l1 = 0, lambda_offset = 0,
  strides = c(1L, 1L), padding = "valid", activation = "sigmoid",
  use_bias = TRUE, kernel_initializer = initializer_random_uniform(0,
  0.5), bias_initializer = "zeros",
  kernel_regularizer = total_regularizer,
  kernel_constraint = info_constraint,
  bias_constraint = negative_constraint, input_shape = NULL,
  name = "deNovo_conv", trainable = NULL)

`object`	Keras model object.
`filters`	Number of convolutional filters.
`filter_len`	Length of the motifs.
`lambda_pos`	Position-wise L1 sparsity penalty.
`lambda_filter`	Sparse group lasso penalty.
`lambda_l1`	Sparse L1 penalty applied to filter weights.
`lambda_offset`	L1 penalty applied the offsets (bias) terms.
`strides`	An integer or list of 2 integers, specifying the strides of the convolution along the width and height. Can be a single integer to specify the same value for all spatial dimensions. Specifying any stride value != 1 is incompatible with specifying any `dilation_rate` value != 1.
`padding`	one of "valid" (default) or "same" (case-insensitive).
`activation`	Activation function to use. Default is the sigmoid activation.
`use_bias`	Boolean, whether the layer uses a bias vector. Default is to include a bias.
`kernel_initializer`	Initializer for the `kernel` weights matrix. Default is random uniform on `(0,0.5)` (i.e. `initializer_random_uniform(0,.5)`).
`bias_initializer`	Initializer for the bias vector. Default is `'zeros'`.
`kernel_regularizer`	Regularizer function applied to the `kernel` weights matrix. Options include `NULL`, `total_regularizer` (default), `position_regularizer`, or `filter_regularizer`.
`kernel_constraint`	Constraint function applied to the kernel matrix. Options include `NULL`, `info_constraint` (default), `pwm_constraint`, `negative_constraint`, or `nonnegative_constraint`.
`bias_constraint`	Constraint function applied to the bias vector.Options include `NULL`, `negative_constraint` (default), or `nonnegative_constraint`.
`input_shape`	Dimensionality of the input (integer) not including the samples axis. This argument is required when using this layer as the first layer in a model.
`name`	An optional name string for the layer. Should be unique in a model (do not reuse the same name twice). It will be autogenerated if it isn't provided. Default is `'deNovo_conv'`.
`trainable`	Whether the layer weights will be updated during training. Must be specified.