T31. FEWER TELLS MORE: A POLYGENIC RISK SCORE VARIANT PRIORITIZATION PIPELINE FOR DEPRESSION ACROSS ANCESTRIES
EUROPEAN NEUROPSYCHOPHARMACOLOGY(2024)
Karolinska Inst
Abstract
Background Large-scale genome-wide association studies (GWAS) for depression predominantly target individuals of European population, leading to underpowered depression polygenic risk scores (PRS) and hampering effective risk stratification for non-European populations. Integrating GWAS SNPs with functional annotations has been suggested to enhance PRS predictability. We hypothesized that incorporating functional annotations and brain eQTLs would improve the predictability and generalizability of depression PRS across diverse populations. Methods We developed a variant prioritization pipeline for depression PRS: 1) using fGWAS to screen and identify functional annotations enriched for depression GWAS signals, we priotitized variants based on log Bayes factor (logBF) and posterior probability of association; 2) Expression-trait associations were assessed using brain cis-eQTL (T Qi et al, 2018) and depression GWAS(T.D. Als et al, 2024) summary statistics to prioritize SNPs potentially influencing depression via their effect on gene expression. These steps resulted in three prioritized variant groups: 95% credible set, logBF> 4, and transcriptome-wide significant (TWS). Continuous shrinkage priors were then applied, via PRS-CS, to generate weights from GWAS summary statistics for each variant group and an all-variant benchmark. Using these results, depression PRSs were generated for African American (AA), European (EUR), and Hispanic (HISP) individuals in the Women's Health Initiative study. Logistic regression models were applied on a narrow depression definition adjusting for age, genotyping platform, and ten genetic PCs. Predictive performance of PRSs was evaluated in each population using adjusted R2, AUC, and odds ratio with 95% confidence intervals. Results The 95% credible set PRS (∼6.2% of all variants) improved prediction accuracy by 67.6% in EUR, 181% in AA, and 656% in HISP compared to all-variant benchmark PRS. It has the greatest increase in AUC compared to all the other PRS across all ancestries, with EUR showing the highest increase. Moreover, the 95% credible set PRS was the only PRS that significantly predicted increased risk of depression across all ancestries. The logBF> 4 set PRS (∼1 % of all variants) had the highest per-variant prediction accuracy compared to all other PRS, while TWS PRS (∼1.4 % of all variants) had slightly higher per-variant R2 compared to the benchmark all-variant PRS. However, they did not outperform the 95% credible set. Discussion Integrating biological information enhances PRS robustness within and across ancestries. PRS, using fewer but biologically informed variants, captures essential information effectively. Larger eQTL datasets that include cell type results are needed to improve the power of TWS variant prioritization.
MoreTranslated text
求助PDF
上传PDF
View via Publisher
AI Read Science
AI Summary
AI Summary is the key point extracted automatically understanding the full text of the paper, including the background, methods, results, conclusions, icons and other key content, so that you can get the outline of the paper at a glance.
Example
Background
Key content
Introduction
Methods
Results
Related work
Fund
Key content
- Pretraining has recently greatly promoted the development of natural language processing (NLP)
- We show that M6 outperforms the baselines in multimodal downstream tasks, and the large M6 with 10 parameters can reach a better performance
- We propose a method called M6 that is able to process information of multiple modalities and perform both single-modal and cross-modal understanding and generation
- The model is scaled to large model with 10 billion parameters with sophisticated deployment, and the 10 -parameter M6-large is the largest pretrained model in Chinese
- Experimental results show that our proposed M6 outperforms the baseline in a number of downstream tasks concerning both single modality and multiple modalities We will continue the pretraining of extremely large models by increasing data to explore the limit of its performance
Upload PDF to Generate Summary
Must-Reading Tree
Example

Generate MRT to find the research sequence of this paper
Data Disclaimer
The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn
Chat Paper
Summary is being generated by the instructions you defined