Related Experiment Video
Updated: Nov 15, 2025
09:06
Cryo-EM and Single-Particle Analysis with Scipion
Published on: May 29, 2021
4.2K
R.ROSETTA: an interpretable machine learning framework
Mateusz Garbulowski1, Klev Diamanti1,2, Karolina Smolińska1
1Department of Cell and Molecular Biology, Uppsala University, Uppsala, Sweden.
BMC Bioinformatics
|March 7, 2021
Summary
This study introduces R.ROSETTA, an interpretable machine learning package using rough set theory. It aids bioinformatics by providing transparent models and statistical insights, particularly for analyzing gene dependencies in autism research.
Area of Science:
- Bioinformatics
- Computational Biology
- Machine Learning
Background:
- Machine learning offers data mining and knowledge discovery in bioinformatics.
- Interpretable machine learning is crucial for understanding prediction mechanisms in Life Sciences.
- Rough set theory provides a foundation for developing interpretable models.
Purpose of the Study:
- Implement an interpretable machine learning package based on rough set theory.
- Provide statistical properties of the developed models and their components.
- Facilitate transparent and accessible results for the scientific community.
Main Methods:
- Developed R.ROSETTA, an R wrapper for the ROSETTA framework.
- Implemented rule-based modeling and combinatorial statistics.
- Integrated statistics and visualization tools to minimize analysis bias and noise.
Main Results:
- R.ROSETTA enables building and analyzing non-linear interpretable machine learning models.
- Applied the package to a transcriptome dataset from an autism case-control study.
- Identified potential co-predictive mechanisms among neurodevelopmental and autism-related genes.
Conclusions:
- R.ROSETTA offers novel insights for interpretable machine learning and knowledge-based systems.
- The package successfully detected dependencies among autism-related genes.
- R.ROSETTA is versatile and applicable to any data organized in decision tables.
