Search

BackgroundHigh-dimensional clinical data are becoming more accessible in biobank-scale datasets. However, accurately phenotyping high-dimensional clinical data remains a major impediment to genetic discovery.MethodsWe introduce a general deep learning framework, RE presentation learning for Genetic discovery on Low-dimensional Embeddings (REGLE), for discovering associations between genetic variants and high-dimensional clinical data. REGLE uses convolutional variational autoencoders to compute anon-linear, low-dimensional, disentangled embeddingof the data and can also incorporate expert clinical metrics. We demonstrate the utility of REGLE by application to spirograms, which measure lung function. We generate two types of synthetic representations of pulmonary functions we call spirogram encodings (SPINCs) and residual spirogram encodings (RSPINCs).FindingsGenome-wide association studies on (R)SPINCs identify more genome-wide significant loci than existing methods while replicating most known lung function loci. Furthermore, (R)SPINCs are associated with overall survival and, under the latent causal variable model, they exhibit significantly high genetic causality proportion with asthma, chronic obstructive pulmonary disease (COPD), and inflammatory diseases. Finally, we construct a set of polygenic risk scores (PRS) that are generally predictive of pulmonary traits and diseases. We demonstrate superior performance predicting asthma and COPD, in multiple ancestries and across four biobanks, compared to PRSs constructed using expert-defined pulmonary function measurements.InterpretationREGLE is a method for generating low-dimensional, disentangled representations of high-dimensional clinical data that does not require labels, and improves upon expert-defined phenotypes for genetic discovery and disease prediction. It can flexibly incorporate expert-defined or clinical features and provides a framework to create accurate disease-specific PRS in datasets which have minimal expert phenotyping. (R)SPINCs are quantifying clinically relevant features that are not currently captured in a standardized or automated way.FundingGoogle LLC.

Genome-wide association studies (GWAS) are used to identify genetic variants significantly correlated with a target disease or phenotype as a first step to detect potentiallycausalgenes. The availability of high-dimensional biomedical data in population-scale biobanks has enabled novel machine-learning-based phenotyping approaches in which machine learning (ML) algorithms rapidly and accurately phenotype large cohorts with both genomic and clinical data, increasing the statistical power to detect variants associated with a given phenotype. While recent work has demonstrated that these methods can be extended to diseases for which only low quality medical-record-based labels are available, it is not possible to quantify changes in statistical power since the underlying ground-truth liability scores for the complex, polygenic diseases represented by these medical-record-based phenotypes is unknown. In this work, we aim to empirically study the robustness of ML-based phenotyping procedures to label noise by applying varying levels of random noise to vertical cup-to-disc ratio (VCDR), a quantitative feature of the optic nerve that is predictable from color fundus imagery and strongly influences glaucoma referral risk. We show that the ML-based phenotyping procedure recovers the underlying liability score across noise levels, significantly improving genetic discovery and PRS predictive power relative to noisy equivalents. Furthermore, initial denoising experiments show promising preliminary results, suggesting that improving such methods will yield additional gains.

Chronic obstructive pulmonary disease (COPD), the third leading cause of death worldwide, is highly heritable. While COPD is clinically defined by applying thresholds to summary measures of lung function, a quantitative liability score has more power to identify new genetic signals. Here we train a deep convolutional neural network on noisy self-reported and ICD-based labels to predict COPD case/control status from high-dimensional raw spirograms and use the model predictions as a liability score. The machine-learning-based (ML-based) liability score accurately discriminates COPD cases and controls (AUROC = 0.82 ± 0.01) and COPD-related hospitalization (AUROC = 0.89 ± 0.01) without any domain-specific knowledge. Moreover, the ML-based liability score is associated with overall survival (Hazard ratio = 1.22 ± 0.01; P ≤ 2 × 10−16) and exacerbation events (R2 = 0.10 ± 0.01; P ≤ 4 × 10−101). A genome-wide association study on the ML-based liability score replicates existing COPD and lung function loci, but also identifies 67 new loci. Thirty-eight of these have supportive evidence in independent datasets, including a locus near LTBR. We demonstrate the biological plausibility of the novel variants through enrichment analyses, phenome-wide association studies, and generalizability of COPD prediction in multiple datasets. These results provide an example of the potential to improve genetic discovery of disease-relevant variants by training deep neural networks to predict noisy labels from high-dimensional raw data.

Summary Genome-wide association studies (GWASs) require accurate cohort phenotyping, but expert labeling can be costly, time intensive, and variable. Here, we develop a machine learning (ML) model to predict glaucomatous optic nerve head features from color fundus photographs. We used the model to predict vertical cup-to-disc ratio (VCDR), a diagnostic parameter and cardinal endophenotype for glaucoma, in 65,680 Europeans in the UK Biobank (UKB). A GWAS of ML-based VCDR identified 299 independent genome-wide significant (GWS; p ≤ 5 × 10−8) hits in 156 loci. The ML-based GWAS replicated 62 of 65 GWS loci from a recent VCDR GWAS in the UKB for which two ophthalmologists manually labeled images for 67,040 Europeans. The ML-based GWAS also identified 93 novel loci, significantly expanding our understanding of the genetic etiologies of glaucoma and VCDR. Pathway analyses support the biological significance of the novel hits to VCDR: select loci near genes involved in neuronal and synaptic biology or harboring variants are known to cause severe Mendelian ophthalmic disease. Finally, the ML-based GWAS results significantly improve polygenic prediction of VCDR and primary open-angle glaucoma in the independent EPIC-Norfolk cohort.