API¶

You can import perturbseq as:

import perturbseq as perturb

Note

Testnote

Preprocessing: `pp`¶

Perturbation assignments and queries

`pp.compute_TPT`(gbcs_dataset)	Compute transcript per transcript
`pp.get_perturbations`(adata_here[, …])	Get a list of perturbations in the dataset
`pp.perturb_overlap_obs`(perturbation_list, …)	Get perturbations present as obs
`pp.annotate_controls`(adata_here, …[, …])	Make obs with control guides
`pp.subset_singly_perturbed`(adata_here[, …])	Keep only cells with one perturbation
`pp.perturbs_per_cell`(adata_here[, …])	Number of perturbations per cell
`pp.cells_per_perturb`(adata_here[, …])	Counts the number of cells for each perturbation.
`pp.delete_guides_from_varnames`(adata_here[, …])	Delete perturbations from adata.var_names

Sub/down-sampling

`pp.subsample_cells`(adata_here, num_cells, …)	Subsample cells/perturbation
`pp.downsample_counts`(adata_here, …[, my_rng])	Downsample counts per cell to a fraction

Summarization across perturbations

`pp.obs_mean`(adata_here, grouping_variable, obs)	Get the mean of an obs across pre-defined groups.
`pp.score_programs`(adata_here[, …])	Get program score for each cell, by averaging program genes

Preparing for modeling

`pp.obs_to_design_matrix`(adata_here, obs_names)	Design matrix for linear model from obs
`pp.split_train_valid_test`(adata_here[, …])	Split cells into training, validation and test

tl.moi(adata_here[, perturbations_obs, …])