Improving genetic risk prediction across diverse population by disentangling ancestry representations

Trait imputation enhances nonlinear genetic prediction for some traits

The expansive collection of genetic and phenotypic data within biobanks offers an unprecedented opportunity for biomedical research. However, the frequent occurrence of missing phenotypes presents a significant barrier to fully leveraging this potential. In our target application, on one hand, we have only a small and complete dataset with both genotypes and phenotypes to build a genetic prediction model, commonly called a polygenic (risk) score (PGS or PRS); on the other hand, we have a large dataset of genotypes (e.g. from a biobank) without the phenotype of interest. Our goal is to leverage the large dataset of genotypes (but without the phenotype) and a separate genome-wide association studies summary dataset of the phenotype to impute the phenotypes, which are then used as an individual-level dataset, along with the small complete dataset, to build a nonlinear model as PGS. More specifically, we trained some nonlinear models to 7 imputed and observed phenotypes from the UK Biobank data. We then trained an ensemble model to integrate these models for each trait, resulting in higher R2 values in prediction than using only the small complete (observed) dataset. Additionally, for 2 of the 7 traits, we observed that the nonlinear model trained with the imputed traits had higher R2 than using the imputed traits directly as the PGS, while for the remaining 5 traits, no improvement was found. These findings demonstrate the potential of leveraging existing genetic data and accounting for nonlinear genetic relationships to improve prediction accuracy for some traits.

Highly parameterized polygenic scores tend to overfit to population stratification via random effects

Polygenic scores (PGSs), increasingly used in clinical settings, frequently include many genetic variants, with performance typically peaking at thousands of variants. Such highly parameterized PGSs often include variants that do not pass a genome-wide significance threshold. We propose a mathematical perspective that renders the effects of many of these non-significant variants random rather than causal, with the randomness capturing population structure. We devise methods to assess variant effect randomness and population stratification bias. Applying these methods to 141 traits from the UK Biobank, we find that, for many PGSs, the effects of non-significant variants are considerably random, with the extent of randomness associated with the degree of overfitting to population structure of the discovery cohort. Our findings explain why highly parameterized PGSs simultaneously have superior cohort-specific performance and limited generalizability, suggesting the critical need for variant randomness tests in PGS evaluation. Supporting code and a dashboard are available at https://github.com/songlab-cal/StratPGS.

Validation of a Community-Based Approach Toward Personalized Dementia Risk Reduction: The Kimel Family Centre for Brain Health and Wellness

BACKGROUND/OBJECTIVES

The Kimel Family Centre for Brain Health and Wellness is a research-driven community centre testing the efficacy of personalized dementia risk reduction programming on dementia risk and cognition. The objective of this protocol is to validate this approach by following people for two years.

DESIGN/SETTING

Participants will receive a comprehensive dementia risk assessment, including nonmodifiable and modifiable risk factors, from which they will receive a Personalized Dementia Risk Report and Program Strategy, indicating their health conditions increasing and their risk level in five modifiable risk domains: physical activity, brain-healthy eating, cognitive engagement, social connections, and mental wellbeing. Equipped with this information, participants will enroll in programs within the Centre to address their risk factors. Changes to their dementia risk, cognition, and Personalized Program Strategy will be communicated through re-assessments of risk factors every six months (risk and cognition) and every year (comprehensive assessment).

PARTICIPANTS

Participants (n = 450) will be 50 years of age or older, without a diagnosis of dementia, and sufficiently fluent in English to complete the assessments and understand program instructors. One goal is that our participant sample will include people of low income (with fundraising providing free community centre membership), and from various ethno-racial backgrounds.

INTERVENTION

Participants will select programs to meet their Personalized Program Strategy. For physical activity, they will gradually work toward the Canadian Society for Exercise Physiology guidelines. For brain-healthy eating, they will learn about the Brain Health Food Guide and food label reading, and then take additional programs. For cognitive engagement and mental wellbeing, they will take at least one hour of relevant programming per week. Social connections will be reinforced throughout all programs. All participants will also have access to the Canadian Consortium on Neurodegeneration's CAN-THUMBS Up online, educational program on modifiable dementia risk factors, called Brain Health PRO.

MEASUREMENTS

The comprehensive assessment includes numerous dementia risk factors, but the primary measures are risk in the five domains, health conditions proximal to those five risk domains, and cognition, and how these are affected by adherence and quality of goal-directed future simulation. We hypothesize a reduced risk in the five domains within six months, improvements in health biomarkers within a year, and maintenance of cognition within two years, with these benefits accruing with greater adherence, but only up to a point, at which benefits will plateau, and greater benefits among participants whose goal-directed simulations are more vivid, personally-relevant, achievable, and positive.

CONCLUSIONS

This innovative approach overcomes a number of limitations present in prior multidomain dementia prevention trials. Adapting a preference clinical trial that is embedded in a community centre, where participants have autonomy to choose programs to address their modifiable dementia risk factors, has real-world applicability in the global effort to reduce dementia risk.

FairPRS: adjusting for admixed populations in polygenic risk scores using invariant risk minimization

Polygenic risk scores (PRS) are increasingly used to estimate the personal risk of a trait based on genetics. However, most genomic cohorts are of European populations, with a strong under-representation of non-European groups. Given that PRS poorly transport across racial groups, this has the potential to exacerbate health disparities if used in clinical care. Hence there is a need to generate PRS that perform comparably across ethnic groups. Borrowing from recent advancements in the domain adaption field of machine learning, we propose FairPRS - an Invariant Risk Minimization (IRM) approach for estimating fair PRS or debiasing a pre-computed PRS. We test our method on both a diverse set of synthetic data and real data from the UK Biobank. We show our method can create ancestry-invariant PRS distributions that are both racially unbiased and largely improve phenotype prediction. We hope that FairPRS will contribute to a fairer characterization of patients by genetics rather than by race.