train_dreams_model: Train error model

In JakobPedersenLab/dreams: Deep Read-level Error Model for Sequencing data

Train error model

Description

Train error model

Usage

train_dreams_model(
  training_data,
  layers,
  model_features,
  lr,
  batch_size,
  epochs,
  model_file_path = NULL,
  log_file_path = NULL,
  min_delta = 0,
  patience = 0,
  l2_reg = 0,
  validation_split = 0,
  ctx3_embed_dim = 3
)

Arguments

training_data	data.frame Input training data (Generated from get_training_data())
layers	Numeric vector. Number of nodes in each layer.
model_features	Vector of feature names. Selected features for model training.
lr	Numeric value between 0 and 1. Learning rate.
batch_size	Integer. Batch size.
epochs	Integer. Number of training epochs.
model_file_path	String. Model output file path. Default is NULL.
log_file_path	String. Path to model output log file. Default is NULL.
min_delta	Numeric value between 0 and 1. Minimum delta for early stopping. Default is 0.
patience	Integer. Patience when reaching minimum delta. Default is 0.
l2_reg	Numeric value between 0 and 1. Level of L2 regularization per layer. Default is 0.
validation_split	Numeric value between 0 and 1. Validation split ratio. Default is 0.
ctx3_embed_dim	Integer. Number of dimensions to embed trinucleotide context to. Default is 3.

Value

Trained model in hdf5 format.

See Also

get_training_data() Function for getting training data

Other Train model: train_dreams_model_indels()

JakobPedersenLab/dreams documentation built on Feb. 2, 2024, 3:14 p.m.