Evaluation and Aggregation of Active Module Identification Algorithms

[agitter]

We should see if this benchmarking project has anything relevant for us https://doi.org/10.1101/2025.10.06.680790

Background High-throughput sequencing methods have generated vast amounts of genetic data for candidate gene studies. As a part of the analysis, candidate genes are often analyzed through Gene-Gene interaction (GGI) networks. These networks can become very large, necessitating efficient methods to reduce their complexity. Active Module Identification (AMI) is a common method to analyze GGI networks by identifying enriched subnetworks representing relevant biological processes. Multiple AMI algorithms have been developed for biological datasets, and a comprehensive assessment of these algorithms and a comparative analysis of their behaviors across a variety of use-cases are crucial for their appropriate applications.

Results In this study, we use the Empirical Pipeline (EMP) to evaluate four AMI algorithms – PAPER, DOMINO, FDRnet, and HotNet2 – on their ability to produce context-specific enrichment. When testing the algorithms on four biological datasets, our results reveal that no single algorithm outperforms the others across all datasets. Moreover, the output modules are often dissimilar, suggesting that different algorithms capture complementary biological signals. Our results suggest that a comprehensive analysis requires the aggregation of outputs from multiple algorithms. We propose two methods to this end: a spectral clustering approach for module aggregation, and an algorithm that combines modules with similar network structures called Greedy Conductance-based Merging (GCM).

Conclusions Overall, our results advance our understanding of AMI algorithms and how they should be applied. Tools and workflows developed in this study will facilitate researchers working with AMI algorithms to enhance their analyses. Our code is freely available at https://github.com/LiuJ0/AMI-Benchmark/.

[tristan-f-r]

If we want to do active module/DMM benchmarking in SPRAS, we could try integrating PAPER and FDRnet and their datasets into SPRAS as well. I'll open issues for those and we can decide if we want them.

They also have a result on parameter selection that @ntalluri may be interested in (3.5: Merging algorithms could help overcome parameter selection)

[ntalluri]

We don’t plan to separate active module or DMM benchmarking for the paper. The plans for the paper as of now is to benchmark all algorithms together and treat them as pathway reconstruction methods.

• So there won't be a separate section soley benchmarking active module or DMM algorithms only

[ntalluri]

This paper is good to mention in the related works section.

[ntalluri]

They also do something kind of similar for parameter selection that is like our PCA method and ensembling method except they use spectral clustering.

[agitter]

I noticed the GitHub repo https://github.com/LiuJ0/AMI-Benchmark/ gives a 404. It may still be private.