Related Experiment Video
Updated: Dec 26, 2025
03:08
Using Human Differentially Expressed Gene Lists to Perform Downstream Pathway Enrichment Analysis and Target Prioritization
Published on: October 3, 2025
823
Robust pathway sampling in phenotype prediction. Application to triple negative breast cancer
Ana Cernea1, Juan Luis Fernández-Martínez2, Enrique J deAndrés-Galiana1,3
1Group of Inverse Problems, Optimization and Machine Learning, Department of Mathematics, University of Oviedo, C/ Federico García-Lorca, 18, 33007, Oviedo, Spain.
BMC Bioinformatics
|March 14, 2020
Summary
Three novel sampling algorithms accurately identify defective genetic pathways in triple-negative breast cancer (TNBC) and offer a computationally efficient alternative to Bayesian Networks for phenotype prediction.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Phenotype prediction is challenging due to limited samples and numerous genetic networks.
- Identifying defective genetic pathways is complicated by high dimensionality.
- Triple-negative breast cancer (TNBC) presents complex genetic pathway alterations.
Purpose of the Study:
- To introduce and evaluate three novel sampling algorithms for phenotype prediction.
- To identify, classify, and characterize defective genetic pathways in TNBC.
- To compare the performance of novel samplers against Bayesian Networks.
Main Methods:
- Developed and applied Fisher's ratio, Holdout, and Random samplers.
- Utilized gene sampling to identify frequently altered biological pathways.
- Compared results with those obtained using Bayesian Networks (BNs).
Main Results:
- The novel samplers demonstrated higher accuracy and robustness than BNs.
- Samplers provided comparable insights into disease genomics.
- Biological invariance was confirmed, with pathways independent of sampling method.
Conclusions:
- The tested samplers are effective and less computationally intensive alternatives to BNs.
- The study supports the concept of biological invariance in pathway analysis.
- Bayesian Networks require modifications for accurate uncertainty space sampling in phenotype prediction.
