gemini

GEMINI models sample-dependent and independent effects using log-fold changes and infers these effects through a variational Bayes approach to score and identify genetic interactions such as lethality and recovery.

Key Features:

• Effect modeling: GEMINI uses log-fold changes to model both sample-dependent and sample-independent effects from pairwise CRISPR screen data.
• Variational inference: GEMINI applies a variational Bayes approach to infer the modeled effects.
• Genetic interaction scoring: The inferred effects are used to score and identify genetic interactions, including lethality and recovery phenotypes.

Scientific Applications:

• Genetic interaction identification: GEMINI identifies genetic interactions from pairwise CRISPR screens, enabling detection of synthetic lethal and recovery relationships between genes.

Methodology:

GEMINI uses log-fold changes to model sample-dependent and independent effects, and uses a variational Bayes approach to infer these effects; the inferred effects are then used to score and identify genetic interactions.