Command-line interface

bulkDGD is structured as an importable Python package.

However, a command-line interface is provided for some of the most common tasks bulkDGD is used for.

This interface consists of a series of executables installed together with the package:

• bulkdgd get recount3 allows the seamless retrieval of RNA-seq data and their associated metadata from the Recount3 platform.

• bulkdgd get genes allows the creation of custom lists of genes to use with the bulkDGD model.

• bulkdgd preprocess samples allows the preprocessing of samples’ data before using them with the bulkDGD model.

• bulkdgd find representations allows finding the best representations in the latent space defined by the bulkDGD model for a set of new samples.

• bulkdgd dea allows performing differential gene expression analysis between a set of samples and their ‘normal’ counterparts found by the bulkDGD model.

• bulkdgd reduction allows performing dimensionality reduction analyses and plotting the results.

• bulkdgd find probability_density allows finding, for a given a set of representations, the probability density of each representation for each component of the Gaussian mixture model that defines the bulkDGD model’s latent space.

• bulkdgd train allows training the bulkDGD model.