A fast lasso-based method for inferring higher-order interactions

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Large-scale genotype-phenotype screens are crucial for identifying molecular alterations linked to phenotypes.
• Epistatic effects, particularly gene-gene interactions, are vital in association studies and have implications in areas like antimicrobial resistance.
• Current tools for exome-wide screens primarily focus on pairwise interactions, lacking the capacity to analyze three-way interactions.

Purpose of the Study:

• To develop and validate a novel computational method, Pint, capable of analyzing three-way gene interactions in large-scale genetic datasets.
• To improve the performance and scalability of methods for detecting complex genetic interactions.
• To apply the developed method to real-world biological data, including antibiotic resistance and siRNA perturbation screens.

Main Methods:

• Development of the Pint algorithm, an extension of state-of-the-art methods to incorporate three-way interaction analysis.
• Application of Pint to simulated datasets to assess its performance against existing methods.
• Validation of Pint on real-world datasets from antibiotic resistance testing and siRNA perturbation screens.

Main Results:

• Pint demonstrated superior performance compared to known methods on simulated data.
• The method successfully identified biologically plausible gene effects in both antibiotic resistance and siRNA perturbation models.
• A specific combination of known tumor suppressor genes was identified by Pint as potentially increasing cell proliferation.

Conclusions:

• Pint offers a significant advancement in the analysis of large-scale genetic data by enabling the detection of three-way interactions.
• The method has broad applicability in fields ranging from microbial genetics to human disease research.
• Pint provides a powerful tool for uncovering complex genetic architectures underlying various biological phenotypes.