Home

Awesome

Anchor logo

codecov

What is anchor?

Anchor is a python package to find unimodal, bimodal, and multimodal features in any data that is normalized between 0 and 1, for example alternative splicing or other percent-based units.

Installation

To install anchor, we recommend using the Anaconda Python Distribution and creating an environment, so the anchor code and dependencies don't interfere with anything else. Here is the command to create an environment:

conda create -n anchor-env pandas scipy numpy matplotlib seaborn

Stable (recommended)

To install this code from the Python Package Index, you'll need to specify anchor-bio (anchor was already taken - boo).

pip install anchor-bio

Bleeding-edge (for the brave)

If you want the latest and greatest version, clone this github repository and use pip to install

git clone git@github.com:YeoLab/anchor
cd anchor
pip install .  # The "." means "install *this*, the folder where I am now"

Usage

anchor was structured like scikit-learn, where if you want the "final answer" of your estimator, you use fit_transform(), but if you want to see the intermediates, you use fit().

If you want the modality assignments for your data, first make sure that you have a pandas.DataFrame, here it is called data, in the format (samples, features). This uses a log2 Bayes Factor cutoff of 5, and the default Beta distribution parameterizations (shown here)

import anchor

bm = anchor.BayesianModalities()
modalities = bm.fit_transform(data)

If you want to see all the intermediate Bayes factors, then you can do:

import anchor

bm = anchor.BayesianModalities()
bayes_factors = bm.fit(data)

History

1.1.1 (2017-06-29)

1.0.1 (2017-06-28)

1.0.0 (2017-06-28)

0.1.0 (2015-07-08)