isGWAS: ultra-high-throughput, scalable and equitable inference of genetic associations with disease

Objective:

The aim of this research was to develop scalable and efficient algorithms, namely the in-silico GWAS (isGWAS) and leapfrog re-sampler (LRS), to overcome significant challenges in genome-wide association studies (GWAS). These challenges include the dependency on sensitive individual-level data, daunting computational demands arising from increasing cohort sizes and genetic variants, and limitations in traditional regression-based methodologies. Moreover, the objective was to democratize GWAS by providing user-friendly interfaces, thereby revolutionizing the approach to genetic research.

Solutions:

• Development of the in-silico GWAS (isGWAS) algorithm by Optima and bioXcelerate AI, capable of deducing regression parameters from cohort-level summary data.
• Creation of the leapfrog re-sampler (LRS) algorithm by bioXcelerate AI to enhance association result projection onto semi-virtually expanded cohorts.
• Incorporation of user-friendly web interfaces to facilitate customizable usage and accessibility.

Challenges:

• Dependence on sensitive individual-level data, posing privacy concerns.
• Daunting computational requirements due to increasing cohort sizes, genetic variants, and diverse phenotypes.
• Pressure on algorithms, leading to computational bottlenecks.
• Ensuring accuracy and reliability of results amidst computational challenges.
• Accessibility in resource-constrained environments to democratize genetic research.

Impact:

• Unparalleled scalability and computing prowess of isGWAS, enabling rapid deduction of regression parameters and processing variants 1500 times faster than previous technology.
• Results from isGWAS comparable in accuracy and reliability to traditional regression-based methodologies.
• Democratization of GWAS through user-friendly interfaces, making genetic research more accessible, particularly in resource-constrained settings.
• A paradigm shift in genetic research, redefining the approach to GWAS and opening avenues for deeper exploration of genetic underpinnings in disease and trait associations.
• Highlighting the potential of the ProteinScore for type 2 diabetes, surpassing existing clinical markers and providing insights into the role of proteomic signatures in disease prognosis.