API

Core

DataOptions(*[, group_by, layer, ...])	Options for the data.
ModelOptions(*[, n_factors, weight_prior, ...])	Options for the model.
TrainingOptions(*[, device, batch_size, ...])	Options for training.
SmoothOptions(*[, n_inducing, kernel, ...])	Options for Gaussian processes.
MOFAFLEX(data, *args)	Fit the model using the provided data.
FeatureSet(features, name[, description])	Class for storing a single set of features (genes).
FeatureSets(feature_sets[, name, remove_empty])	Class for storing a collection of feature sets (see FeatureSet).

Presets

Presets corresponding to different previously published factor analysis models.

These can be used by passing them to the MOFAFLEX constructor: MOFAFLEX(*preset).

mofaflex.presets.MOFA

Options used to reproduce MOFA results in the MOFA-FLEX paper.

mofaflex.presets.MEFISTO

Options used to reproduce MEFISTO results in the MOFA-FLEX paper.

mofaflex.presets.NSF

Options used to reproduce NSF results in the MOFA-FLEX paper.

Settings

An instance of the _core.settings.Settings class is available as mofaflex.settings and allows configuring MOFA-FLEX.

_core.settings.Settings(*[, use_dask, ...])

Global settings.

Tools

tl.DataGenerator(n_features[, n_samples, ...])	Generator class for creating synthetic multi-view data with latent factors.
tl.factor_correlation(model)	Calculate the correlation between factors.
tl.get_free_gpu_idx()	Get the index of the GPU with current lowest memory usage.
tl.match(reference, permutable, axis)	Find optimal permutation and signs to match two tensors along specified axis.
tl.msigdb_get_features([category, dbver, entrez])	Get gene sets from the MSIGDB molecular signatures database.
tl.msigdb_list_categories([dbver])	List available MSIGDB categories for the given database version.
tl.msigdb_list_versions()	List available versions of the MSIGDB molecular signature database.
tl.test_annotation_significance(model, annotations)	Test feature sets for significant associations with model factors.

Plotting

pl.all_weights(model[, views, clip, ...])	Plot the weight matrices.
pl.covariates_factor_scatter(model, factor[, ...])	Plot a factor against one or two covariate dimensions.
pl.factor(model[, factor, show_featurenames, ...])	Plot factor values (y-axis) for each sample (x-axis).
pl.factor_correlation(model[, figsize])	Plot the correlation between factors.
pl.factor_significance(model[, n_factors, ...])	Plot an overview of the factors summarizing the PCGSE results along with the variance explained per factor.
pl.factor_sparsity_histogram(model[, bins, ...])	Plot a histogram of probabilities that factors are non-zero for views with SnS prior.
pl.factors_covariate(model, covariate1[, ...])	Plot every factor against one or two covariates.
pl.factors_scatter(model, x, y[, groups, ...])	Plot two factors against each other and color by covariates.
pl.gp_covariate(model[, ci_opacity, group, ...])	Plot the fitted GP mean for each factor in each group at the data covariate locations.
pl.overview(data[, group_by, missingcolor, ...])	Generate an overview plot of missing data across different views and groups.
pl.smoothness(model[, figsize])	Plot the smoothness of the GP for each factor.
pl.top_weights(model[, n_features, views, ...])	Plot the top weights for a given factor and view.
pl.training_curve(model[, linecolor, ...])	Plot the training curve: -ELBO vs epoch.
pl.variance_explained(model[, group_by, figsize])	Plot the variance explained per factor in each group and view.
pl.weight_sparsity_histogram(model[, bins, ...])	Plot a histogram of probabilities that weights are non-zero for views with SnS prior.
pl.weights(model[, n_features, views, factors, ...])	Plot the weights for a given factor and view.