Command Line Interface

seffnet

Side Effects Knowledge Graph Embeddings.

seffnet [OPTIONS] COMMAND [ARGS]...

optimize

Run the optimization pipeline for a given method and graph.

seffnet optimize [OPTIONS]

Options

--input-path <input_path>

Input graph file. Only accepted edgelist format.

--training-path <training_path>

training graph file. Only accepted edgelist format.

--testing-path <testing_path>

testing graph file. Only accepted edgelist format.

--method <method>

The NRL method to train the model [required]

Options

node2vec|DeepWalk|HOPE|GraRep|LINE|SDNE

--seed <seed>

--prediction-task <prediction_task>

The prediction task for the model [required]

Options

link_prediction|node_classification

--labels-file <labels_file>

The labels file for node classification

--trials <trials>

the number of trials done to optimize hyperparameters

--dimensions-range <dimensions_range>

the range of dimensions to be optimized

--storage <storage>

SQL connection string for study database. Example: sqlite:///optuna.db

--name <name>

Name for the study

-o, --output <output>

Output study summary

--weighted

True if graph is weighted.

--classifier-type <classifier_type>

Choose type of classifier for predictive model

Options

LR|EN|SVM|RF|ENCV

predict

Predict for a given entity.

seffnet predict [OPTIONS] CURIE

Options

-n, --number-predictions <number_predictions>

-t, --result-type <result_type>

Options

chemical|phenotype|target

Arguments

CURIE

Required argument

predictc

Predict for a chemical by SMILES string.

seffnet predictc [OPTIONS] SMILES

Options

-n, --number-predictions <number_predictions>

-t, --result-type <result_type>

Options

chemical|phenotype|target

Arguments

SMILES

Required argument

rebuild

Build all resources from scratch.

seffnet rebuild [OPTIONS]

repeat

Repeat training n times.

seffnet repeat [OPTIONS]

Options

--input-path <input_path>

Input graph file. Only accepted edgelist format.

--training-path <training_path>

training graph file. Only accepted edgelist format.

--testing-path <testing_path>

testing graph file. Only accepted edgelist format.

--method <method>

The NRL method to train the model [required]

Options

node2vec|DeepWalk|HOPE|GraRep|LINE|SDNE

--evaluation-file <evaluation_file>

The path to save evaluation results.

--dimensions <dimensions>

The dimensions of embeddings.

--number-walks <number_walks>

The number of walks for random-walk methods.

--walk-length <walk_length>

The walk length for random-walk methods.

--window-size <window_size>

The window size for random-walk methods.

--p <p>

The p parameter for node2vec.

--q <q>

The q parameter for node2vec.

--alpha <alpha>

The alpha parameter for SDNE

--beta <beta>

The beta parameter for SDNE

--epochs <epochs>

The epochs for deep learning methods

--kstep <kstep>

The kstep parameter for GraRep

--order <order>

The order parameter for LINE. Could be 1, 2 or 3

--n <n>

number of repeats.

--seed <seed>

--weighted

True if graph is weighted.

--prediction-task <prediction_task>

The prediction task for the model [required]

Options

link_prediction|node_classification

--classifier-type <classifier_type>

Choose type of classifier for predictive model

Options

LR|EN|SVM|RF|ENCV

--randomization <randomization>

Options

xswap|random|node_shuffle

train

Train my model.

seffnet train [OPTIONS]

Options

--input-path <input_path>

Input graph file. Only accepted edgelist format.

--training-path <training_path>

training graph file. Only accepted edgelist format.

--testing-path <testing_path>

testing graph file. Only accepted edgelist format.

--seed <seed>

--method <method>

The NRL method to train the model [required]

Options

node2vec|DeepWalk|HOPE|GraRep|LINE|SDNE

--evaluation

If true, a testing set will be used to evaluate model.

--evaluation-file <evaluation_file>

The path to save evaluation results.

--embeddings-path <embeddings_path>

The path to save the embeddings file

--predictive-model-path <predictive_model_path>

The path to save the prediction model

--training-model-path <training_model_path>

The path to save the model used for training

--dimensions <dimensions>

The dimensions of embeddings.

--number-walks <number_walks>

The number of walks for random-walk methods.

--walk-length <walk_length>

The walk length for random-walk methods.

--window-size <window_size>

The window size for random-walk methods.

--p <p>

The p parameter for node2vec.

--q <q>

The q parameter for node2vec.

--alpha <alpha>

The alpha parameter for SDNE

--beta <beta>

The beta parameter for SDNE

--epochs <epochs>

The epochs for deep learning methods

--kstep <kstep>

The kstep parameter for GraRep

--order <order>

The order parameter for LINE. Could be 1, 2 or 3

--classifier-type <classifier_type>

Choose type of classifier for predictive model

Options

LR|EN|SVM|RF|ENCV

--weighted

True if graph is weighted.

--prediction-task <prediction_task>

The prediction task for the model [required]

Options

link_prediction|node_classification

--labels-file <labels_file>

The labels file for node classification

update

Update node2vec training model.

seffnet update [OPTIONS]

Options

--updated-graph <updated_graph>

an edgelist containing the graph with new nodes

--chemicals-list <chemicals_list>

a file containing list of chemicals to update the model with

--old-graph <old_graph>

The graph needed to be updated. In pickle format

--updated-graph-path <updated_graph_path>

The path to save the updated fullgraph [required]

--chemsim-graph-path <chemsim_graph_path>

The path to save the chemical similarity graph [required]

--training-model-path <training_model_path>

The path to save the model used for training [required]

--new-training-model-path <new_training_model_path>

the path of the updated training model [required]

--embeddings-path <embeddings_path>

The path to save the embeddings file

--predictive-model-path <predictive_model_path>

The path to save the prediction model

--seed <seed>

web

Run the RESTful API.

seffnet web [OPTIONS]

Options

--host <host>

--port <port>