Autoplier

Embed Omics data using an interpretable Tensorflow based model.

Overview

This package contains a Tensorflow model built in spirit of the PLIER ¹ approach to embed Omics data into a latent space representation.

Installation

pip install -U git+https://github.com/dmontemayor/jdrfmetabo.git@devel

Issues and Bug Reports

Please open an issue if you find any errors or unexpected behavior. Please make sure to document:

1. What you tried
2. What the result was
3. What you expected the result to be
4. Steps (if any) which you took to resolve the issue and their outcomes.

Requirements

The following packages are required to use Autoplier:

• numpy²
• pandas³
• sklearn⁴
• tensorflow⁵

Usage

Setting seed

It is recommended to set a seed when using this package for reproducability. This can be done by calling autoplier.model.set_seed_(seed)

Create model

A model can be created by calling the constructor autoPLIER() from autoplier.model.

Parameters:

• n_inputs: dimensionality of your input data(number of samples)
• n_components: dimensionality of the latent space representation (number of latent variables, default 100)
• dropout_rate: dropout rate of the model, default 0.09
• regval: regularization value, default 1.20E-3
• alpha_init: default 0.05

Fit

After the model was created it can be fit using autoplier.model.fit()

Parameters:

• x_train: training Dataset used to train the Model.
• callbacks: callbacks for the machine learning Model. default none
• batch_size: default 50
• maxepoch: default 2000
• verbose: default 2
• valfrac: default 0.3

Transform

Transform new data into the latent space representation. Before transformation an encoder needs to be created using autoplier.model.build_encoder(). After an encoder was built new data can be transformed using autoplier.model.transform().

Parameters:

• x_predict: Dataset to be transformed into the latent space representation.
• index: name of the samples used in x_predict.

Citations

1. Jaclyn N. Taroni, Peter C. Grayson, Qiwen Hu, Sean Eddy, Matthias Kretzler, Peter A. Merkel, Casey S. Greene, MultiPLIER: A Transfer Learning Framework for Transcriptomics Reveals Systemic Features of Rare Disease, Cell Systems, Volume 8, Issue 5, 2019, Pages 380-394.e4, https://doi.org/10.1016/j.cels.2019.04.003.

2. Harris, C.R., Millman, K.J., van der Walt, S.J. et al. Array programming with NumPy. Nature 585, 2020, 357–362, DOI: 0.1038/s41586-020-2649-2.

3. Jeff Reback, Wes McKinney, jbrockmendel, Joris Van den Bossche, Tom Augspurger, Phillip Cloud, … Mortada Mehyar, pandas-dev/pandas: Pandas 1.0.3 (Version v1.0.3), 2020, http://doi.org/10.5281/zenodo.3715232

4. Pedregosa et al., Scikit-learn: Machine Learning in Python, JMLR 12, 2011, Pages 2825-2830

5. Martín Abadi et. al, TensorFlow: Large-scale machine learning on heterogeneous systems, 2015. Software available from tensorflow.org.