There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | FARID RAJABLI |
| Institution: | University of Miami |
| Department: | Hussman Institute for Human Genomics |
| Country: | |
| Proposed Analysis: | Title: Estimating Bias-free Polygenic Risk Scores for Alzheimer Disease in Admixed Populations Introduction: Polygenic risk scores (PRSs) have proven a valuable tool in predicting the genetic liability of disease on the basis of genotype data, and may improve screening and prevention strategies [1]. A PRS is typically a cumulative risk derived from the trait-associated variants at a defined significance threshold. Mainly the prediction accuracy relies on the marker selection and weight estimation procedures [2]. However, current studies in admixed populations are heavily biased since (i) they don’t account for the complex admixture linkage disequilibrium structure which limits prediction accuracy in marker selection, and (ii) they use summary statistics generated based primarily on European-descent populations. Consequently, PRS studies in European-descent populations have performed much better than those in admixed populations [3-5]. One possible solution to generate bias-free PRS is to use ancestral methods in admixed populations. We are proposing a new approach in Alzheimer disease (AD) PRS calculation where the PRS construction in admixed populations will be conditioned on local ancestral blocks to generate risk prediction with the higher accuracy. Importantly, bias-free PRS estimations will improve health equality in precision medicine and reduce further health disparities. Analysis to be Performed: Our analysis will focus on (i) developing a novel approach to calculate the polygenic risk scores in admixed populations (ii) by using the ancestry-originated (e.g. European, African (AF), Amerindian (AI)) summary statistics derived through newly developed ancestry-aware association test [6]. We will build a new approach to reduce the biases in constructing PRS in admixed populations. We will create an ancestry-aware PRS weighted by the local ancestry-based statistics using ancestry-originated summary statistics. Then, we will generate also PRSs for each person using summary statistics from non-Hispanic White AD studies, testing both to see if they are significantly different within individuals and to see if they correlate with the overall percent of each ancestry in these individuals. This will give us measure of the difference in the genetic architecture between admixed populations and EU-descent AD. To calculate ancestry-originated summary statistics we will use recently developed ancestry-aware association test. The genetic make-up of admixed populations is not homogenous. Admixture creates mosaic chromosomes of distinct ancestry that lead to a variation of allele frequencies. Due to this structure, in admixed populations more than one ancestral background might be involved across the samples at a certain genomic region. This contributes to the locus-level bias in association studies. We will perform ancestry-aware association analysis to calculate the summary statistics with the certain ancestral background (e.g. AF, AI) by conditioning the test on local ancestry estimates. This will provide ancestry-originated summary statistics that have been lacking, particularly for the AF and AI origin haplotypes. To assess the local ancestry, firstly we will phase dataset by using the Segmented Haplotype Estimation and Imputation tool ver. 2 (ShapeIT) [7]. Then, we will use RFMix [8], discriminative modeling approach, to infer the local ancestry across the genome (we will use reference panels from HGDP and 1000Genome Project.) Association analysis will be performed on stratified set of variants within the certain local ancestral block. The association between PRS and AD will be measured with the area under curve (AUC). We will approximate AUC using SummaryAUC tool [9]. References [1] Purcell, S. M.; et al. (2009). "Common polygenic variation contributes to risk of schizophrenia and bipolar disorder". Nature. 460 (August): 748–752. [2] Chatterjee N, Shi J, Garcia-Closas M. (2016). Developing and evaluating polygenic risk prediction models for stratified disease prevention. Nat. Rev. Genet. 17:392–406 [3] Vilhjálmsson, B. J. et al. (2015) Modeling Linkage Disequilibrium Increases Accuracy of 577 Polygenic Risk Scores. The American Journal of Human Genetics 97, 576–592 578 [4] Belsky, D. W. et al. (2013) Development and Evaluation of a Genetic Risk Score for 587 Obesity. Biodemography and Social Biology 59, 85–100 [5] Wedow, R. et al. (2018) Gene discovery and polygenic prediction from a genome-wide 592 association study of educational attainment in 1.1 million individuals. Nat Genet 593 92, 109 [6] Rajabli, F et al. (2018). Ancestral origin of ApoE epsilon4 Alzheimer disease risk in Puerto Rican and African American populations. PLoS Genet. 12:e1007791. doi: 10.1371/journal.pgen.1007791 [7] Delaneau O, Marchini J. (2014) The 1000 Genomes Project Consortium. Integrating sequence and array data to create an improved 1000 Genomes Project haplotype reference panel. Nature Communications 5:3934. pmid:25653097 [8] Maples BK, Gravel S, Kenny EE, Bustamante CD. (2013) RFMix: a discriminative modeling approach for rapid and robust local-ancestry inference. Am. J. Hum. Genet. 93:278–288 pmid:23910464 [9] Lei S, Aiyi L, Jianxin S. (2019) SummaryAUC: a tool for evaluating the performance of polygenic risk prediction models in validation datasets with only summary level statistics, Bioinformatics, btz176 |
| Additional Investigators |
