Proposing a Method for Inference Gene Regulatory Networks Based on Ensemble Regression and Knowledge Matrix
Test · 22.09.2026
DOI: 10.71500/thesis.2024.290252 ↗Kurz gefasst
Fachartikel in der Zeitschrift Test, erschienen am 22.09.2026. Verfasst von Marzieh Emadi.
Abstract (Originalsprache)
The concept of gene expression and regulatory interactions between genes has been one of the main challenges in recent decades. To understand gene function under various internal and external conditions, it is essential to identify the fundamental processes governing gene function and regulation. In fact, gene expression regulation determines which genes are active and which are inactive at a particular time. Due to the large volume of this data, its analysis requires computational methods for identifying and analyzing gene regulatory networks. However, challenges such as the large number of genes versus the limited number of samples and the presence of noise exist in this field. In this context, the main aim of this research is to present an appropriate modeling or inference method for gene regulatory networks using hybrid methods based on the improved Random Forest and Fuzzy Cognitive Mapping (FCM) based on the Kalman filter. In the first proposed method, regulatory relationships are identified based on the improved Random Forest, while in the second method, regulatory relationships are identified based on the improved Fuzzy Cognitive Map combined with the Kalman filter and compressed sensing. To demonstrate the efficiency of the proposed methods, evaluations are conducted based on various perspectives and dimensions. The criteria reviewed include accuracy, sensitivity, ROC, Ssmean, and others. Additionally, the proposed methods are compared with several common and significant methods in this field. Evaluation results indicate that the presented methods can perform on par with or sometimes better than the significant existing methods in detecting gene regulatory networks. For example, using the DREAM4 dataset, the proposed method achieves an average AUPR of 0.679 for networks of size 10 and an average of 0.259 for networks of size 100, which shows improvement over other methods. The AUPR metric of the proposed method using the IRMA dataset, despite noise in the switch-on state, is 0.963 and 0.713 in the switch-off state. In the absence of noise, the switch-on state achieves 0.928, and the switch-off state achieves 0.748, showing improvement over other methods.
Schlagwörter
Zitieren
APA
Emadi, M. (2026). Proposing a Method for Inference Gene Regulatory Networks Based on Ensemble Regression and Knowledge Matrix. Test. https://doi.org/10.71500/thesis.2024.290252
Vancouver
Emadi M. Proposing a Method for Inference Gene Regulatory Networks Based on Ensemble Regression and Knowledge Matrix. Test. 2026. doi:10.71500/thesis.2024.290252
BibTeX
@article{Emadi2026,
author = {Emadi, Marzieh},
title = {Proposing a Method for Inference Gene Regulatory Networks Based on Ensemble Regression and Knowledge Matrix},
journal = {Test},
year = {2026},
doi = {10.71500/thesis.2024.290252}
}
EUVIMED