ProteoEstimator

• Overview
• Installation
• Usage
  • predict_protein_abundances
  • predict_phospho
• Value
• Note

Overview

We present the first data science competition aiming at predicting protein levels from copy number and transcript levels, as well as phosphorylation levels from protein levels. The winning models outperform standard baseline machine learning methods and simply using the transcript levels as proxy for protein levels with respect to prediction performance on new patient samples. An in depth analysis revealed associations between the commonly predictive genes and essentiality. We provide all the submitted models to the community for re-use and a web application to explore the result of this challenge to support improved large scale proteogenomic characterization of tumor samples and a better understanding of signaling deregulation.

Installation

pip install proteo_estimator

Requires Python3

Usage

import proteo_estimator as pr

# Subchallenge 2: predicting protein levels from copy number and transcript levels
prediction_file_protein = pr.predict_protein_abundances(
            tumor,
            rna,
            cna,
            output_dir,
            logging=True)

# Subchallenge 3: predicting phospho levels from protein abundance and transcript levels
prediction_file_phospho = pr.predict_phospho(
            tumor,
            rna,
            protein,
            output_dir,
            logging=True)

predict_protein_abundances

Arguments

Return Value

predict_phospho

Arguments

Return Value

Note

Please ensure that your docker daemon is running in the background. All file paths must be absolute.