Machine Learning Approaches for the Identification of Genetic Interactions

Area of Science:

• Genetics and Genomics
• Computational Biology
• Bioinformatics

Background:

• Genetic interactions, phenotypic outcomes of gene pair crosstalk, are vital for understanding gene function.
• Synthetic lethality (SL) is a key genetic interaction where inhibiting two genes simultaneously, but not individually, leads to cancer cell death.
• SL interactions are increasingly exploited for targeted cancer therapies.

Purpose of the Study:

• To review machine learning (ML) methods for predicting synthetic lethality (SL) interactions.
• To explain how predicted SL interactions can mediate drug sensitivity in cancer.
• To provide an overview of features used in ML models for SL prediction and their significance.

Main Methods:

• Review of classical machine learning algorithms applied to predict genetic interactions, specifically SL.
• Discussion of feature engineering and selection for training ML models.
• Analysis of the advantages and limitations of various computational approaches.

Main Results:

• Identified various ML algorithms suitable for predicting SL interactions.
• Highlighted the importance of specific genetic and genomic features in model performance.
• Demonstrated the link between SL prediction and potential drug sensitivity outcomes.

Conclusions:

• Machine learning offers powerful tools for identifying synthetic lethality interactions.
• Understanding these interactions and the features driving them is key to developing novel targeted cancer therapies.
• This review provides a foundation for researchers utilizing computational methods in genetic interaction studies.