cpa.ComPertAPI#

class cpa.ComPertAPI(adata, model, de_genes_uns_key='rank_genes_groups_cov', pert_category_key='cov_drug_dose_name', control_group='ctrl', experiment='drug')[source]#

API for CPA model to make it compatible with scanpy.

Methods

`compute_comb_emb`([thrh])	Generates an AnnData object containing all the latent vectors of the cov+dose*pert combinations seen during training.
`compute_uncertainty`(covs, pert, dose[, thrh])	Compute uncertainties for the queried covariate+perturbation combination.
`evaluate_r2`([perturbations, control_adata_key])	Measures different quality metrics about an ComPert autoencoder, when tasked to translate some genes_control into each of the drug/cell_type combinations described in dataset.
`get_covars_embeddings`(covariate)	type covariate: `str`
`get_cycle_uncertainty`(genes_from, df_from, df_to)	Uncertainty for a single condition.
`get_drug_encoding_`(drugs[, doses])	type drugs:
`get_pert_embeddings`([dose])	type dose: default: `1.0`
`get_response`([doses, contvar_min, ...])	Decoded dose response data frame.
`get_response2D`(perturbations, covar[, ...])	Decoded dose response data frame.
`get_response_reference`([perturbations])	Computes reference values of the response.
`latent_dose_response`([perturbations, dose, ...])	type perturbations: default: `None`
`latent_dose_response2D`(perturbations[, ...])	type perturbations:
`mix_drugs`(drugs_list[, doses_list])	Gets a list of drugs combinations to mix, e.g. ['A+B', 'B+C'] and corresponding doses.
`predict`(genes, df[, uncertainty, sample, ...])	Predict values of control 'genes' conditions specified in df.
`print_complete_cycle_uncertainty`(datasets, ...)