Configuration Options (skbio.util.config)#

This module provides a flexible configuration system which allows scikit-bio functions to accept multiple types of input data structures and return results in the users preferred format.

The dispatch system follows a design similar to scikit-learn’s set_output API, allowing users to work with different data formats without changing existing workflows.

Functions#

get_config(option)

Get the current value of an skbio config option.

set_config(option, value)

Set an skbio config option.

Supported Input Formats#

Input handling of supported types is handled automatically. No need to set a configuration variable to use whichever input type you like. When possible, row and column identifiers will be preserved and carried through operations to the results. When IDs are not available from the input data and not provided explicitly, integer indices starting from 0 will be used.

Supported Output Formats#

Configuring Output Format#

There are two ways to control the output format.

The first option is to use the set_config function. This function will change the global behavior of scikit-bio functions.:

# set_config is available as a top level import from skbio
from skbio import set_config

# Set output format to NumPy arrays
set_config("output", "numpy")

# Return to default pandas output
set_config("output", "pandas")

The second option is to set the desired output format on a per-function basis, using the output_format parameter:

from skbio.stats.ordination import cca

# This specific call will return an
# :class:`~skbio.stats.ordination.OrdinationResults` object whose attributes are
# numpy arrays
res = cca(Y, X, output_format="numpy")

Notes#

When using the default pandas backend, existing workflows remain unchanged. Additionally, input and output formats do not have to match.