Package index • deepG

Training

train_model(): Train neural network on genomic data

train_model_cpc(): Train CPC inspired model

resume_training_from_model_card(): Continue training from model card

Model Architecture

create_model_genomenet(): Create GenomeNet Model with Given Architecture Parameters

create_model_lstm_cnn(): Create LSTM/CNN network

create_model_lstm_cnn_multi_input(): Create LSTM/CNN network that can process multiple samples for one target

create_model_lstm_cnn_target_middle(): Create LSTM/CNN network to predict middle part of a sequence

create_model_lstm_cnn_time_dist(): Create LSTM/CNN network for combining multiple sequences

create_model_transformer(): Create transformer model

create_model_twin_network(): Create twin network

remove_add_layers(): Remove layers from model and add dense layers

merge_models(): Merge two models

get_output_activations(): Get activation functions of output layers

reshape_input(): Replace input layer

load_cp(): Load checkpoint

compile_model(): Compile model

layer_aggregate_time_dist_wrapper(): Aggregation layer

layer_pos_embedding_wrapper(): Layer for positional embedding

layer_pos_sinusoid_wrapper(): Layer for positional encoding

layer_transformer_block_wrapper(): Transformer block

Callbacks

conf_matrix_cb(): Confusion matrix callback.

early_stopping_time_cb(): Stop training callback

validation_after_training_cb(): Validation after training callback

reset_states_cb(): Reset states callback

remove_checkpoints(): Remove checkpoints

model_card_cb(): Create model card

Metrics and Losses

noisy_loss_wrapper(): Loss function for label noise

balanced_acc_wrapper(): Balanced accuracy metric

f1_wrapper(): F1 metric

auc_wrapper(): Mean AUC score

exp_decay(): Exponential Decay

stepdecay(): Step Decay

sgdr(): Stochastic Gradient Descent with Warm Restarts

focal_loss_multiclass(): Focal loss for two or more labels

loss_cl(): Contrastive loss

Generators

generator_dummy(): Random data generator

generator_fasta_label_folder(): Data generator for fasta/fasta files

generator_fasta_label_folder_wrapper(): Generator wrapper

generator_fasta_label_header_csv(): Data generator for fasta/fastq files and label targets

generator_fasta_lm(): Language model generator for fasta/fastq files

generator_initialize(): Initializes generators defined by generator_fasta_label_folder function

generator_random(): Randomly select samples from fasta files

generator_rds(): Rds data generator

get_generator(): Wrapper for generator functions

dataset_from_gen(): Collect samples from generator and store in rds or pickle file.

Prediction

predict_model(): Make prediction for nucleotide sequence or entries in fasta/fastq file

summarize_states(): Create summary of predictions

predict_with_n_gram(): Predict the next nucleotide using n-gram

load_prediction(): Read states from h5 file

Evaluation

evaluate_linear(): Evaluate matrices of true targets and predictions from layer with linear activation.

evaluate_model(): Evaluates a trained model on fasta, fastq or rds files

evaluate_sigmoid(): Evaluate matrices of true targets and predictions from layer with sigmoid activation.

evaluate_softmax(): Evaluate matrices of true targets and predictions from layer with softmax activation.

plot_roc(): Plot ROC

Visualization

integrated_gradients(): Compute integrated gradients

heatmaps_integrated_grad(): Heatmap of integrated gradient scores

plot_cm(): Plot confusion matrix

Preprocessing

seq_encoding_label(): Encodes integer sequence for label classification.

seq_encoding_lm(): Encodes integer sequence for language model

get_class_weight(): Estimate frequency of different classes

get_start_ind(): Computes start position of samples

int_to_n_gram(): Encode sequence of integers to sequence of n-gram

n_gram_dist(): Get distribution of n-grams

n_gram_of_matrix(): One-hot encoding matrix to n-gram encoding matrix

reshape_tensor(): Reshape tensors for set learning

split_fasta(): Split fasta file into smaller files.

one_hot_to_seq(): Char sequence corresponding to one-hot matrix.

Data

crispr_sample: CRISPR data

ecoli_small: Ecoli subset

parenthesis: Parenthesis data

Dummy Data

create_dummy_data(): Write random sequences to fasta file

Session Management

deepG deepG-package: deepG for GenomeNet