Bridging Genomics to Phenomics at Atomic Resolution through Variation Spatial Profiling.

Area of Science:

• Genomics
• Computational Biology
• Biophysics

Background:

• Genome sequence variation (genotype) underlies biological diversity and human health (phenotype), including diseases like cystic fibrosis and Alzheimer's.
• Understanding genotype-to-phenotype relationships is crucial for personalized medicine and disease mechanism elucidation.
• Existing methods may not fully capture the complex spatial and temporal dynamics of genetic variation effects.

Purpose of the Study:

• To develop a novel machine learning approach for predicting variant function based on spatial relationships.
• To quantitatively describe the impact of genetic variations on protein structure and function.
• To establish a platform for understanding genotype-to-phenotype transformations in health and disease.

Main Methods:

• Developed a Gaussian-process-based machine learning approach termed variation spatial profiling (VSP).
• Utilized a sparse set of known population variants that perturb protein folds.
• Applied the principle of spatial covariance (SCV) to infer unknown variant functions based on proximity.
• Generated phenotype landscapes using SCV to visualize genotype-to-phenotype relationships.

Main Results:

• Demonstrated that spatial covariance (SCV) quantitatively captures proximity effects in genotype-to-phenotype relationships.
• Showcased the ability of SCV to define functional properties driving polypeptide fold design at atomic resolution.
• Provided proof of principle for using population-based genomic data to understand disease mechanisms.

Conclusions:

• Variation spatial profiling (VSP) offers a novel framework for linking genomic variation to phenotypic outcomes.
• Spatial covariance (SCV) is a powerful principle for modeling the functional impact of genetic variants.
• This approach can advance our understanding of the origins and mechanisms of genotype-to-phenotype transformations in human health and disease.