GEMME

GEMME predicts mutational effects in protein sequences by modeling evolutionary history and inter-site (epistatic) dependencies to generate comprehensive mutational landscapes from sequence alignments.

Key Features:

• Evolutionary Modeling: Explicitly incorporates evolutionary history and models inter-site (epistatic) dependencies, considering all positions within a sequence when estimating mutational effects.
• Efficiency and Speed: Generates a full mutational landscape for a protein from an input alignment in minutes.
• Biologically Meaningful Parameters: Employs a limited number of interpretable, biologically relevant parameters for prediction.
• Versatility Across Protein Families: Validated against 50 high- and low-throughput mutational experiments and shown to predict landscapes for diverse protein families, including viral proteins and highly conserved families.

Scientific Applications:

• Biology, bioengineering, and medicine: Predicts how amino-acid substitutions affect protein function to support basic research, bioengineering projects, and medical studies.
• Disease mechanism analysis: Aids identification of mutations that alter protein function relevant to human disease.
• Therapeutic development: Informs assessment of variant impacts relevant to therapeutic strategies.
• Protein engineering and large-scale studies: Guides engineering of proteins with desired properties and supports analysis of large datasets or high-throughput mutational experiments.

Methodology:

GEMME explicitly models evolutionary history to account for inter-site (epistatic) dependencies, takes an input sequence alignment to compute a full mutational landscape, and uses a limited set of biologically interpretable parameters.