Improved prediction of blood biomarkers using deep learning

Sigurdsson, Arnór I; Ravn, Kirstine; Winther, Ole; Lund, Ole; Brunak, Søren; Vilhjálmsson, Bjarni J.; Rasmussen, Simon

doi:10.1101/2022.10.27.22281549

Cited by 2 publications

(1 citation statement)

References 62 publications

(71 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The copyright holder for this preprint this version posted July 7, 2024. ; https://doi.org/10.1101/2024.07.04.24309942 doi: medRxiv preprint we developed and applied DL frameworks for disease prediction in the UK Biobank 27 , and found potential dominance and epistatic effects, specifically for immunological diseases such as type 1 diabetes (T1D), involving the insulin gene and HLA-DQB1 [28][29][30] . These complex effects also transfer to molecular quantitative trait loci, as indicated by our previous analysis of 34 common biomarkers in the UK Biobank 31 . Additionally, targeted discovery of epistatic effects between genetic variants uncovered the presence of interactions between the ABO blood group and the FUT2 secretor status that influence blood plasma abundance of gastrointestinal (GI) proteins 6 .…”

Section: Introductionmentioning

confidence: 83%

Non-linear genetic regulation of the blood plasma proteome

Sigurdsson,

Gräf,

Yang

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

Although thousands of genetic variants are linked to human traits and diseases, the underlying mechanisms influencing these traits remain largely unexplored. One important aspect is to understand how proteins are regulated by the genome by identifying protein quantitative trait loci (pQTLs). Beyond this, there is a need to understand the role of complex genetics effects such as dominance and epistasis that regulate plasma proteins and protein biomarkers. Therefore, we developed EIR-auto-GP, a deep learning-based approach, to identify such effects. Our results complement the additive genetic regulation identified in previous pQTLs screens by adding a nuanced view of the complex genetic regulation of plasma proteins. Applying this method to the UK Biobank proteomics cohort of 48,594 individuals, we identified 138 proteins that were regulated by non-linear effects, including non-linear covariates (123) as well as genetic dominance and epistasis (15). We uncovered a novel epistatic interaction between theABOandFUT3loci, and demonstrated dominance effects of theABOlocus on plasma levels of pathogen recognition receptors CD209 and CLEC4M. Furthermore, we replicated these findings and the methodology across Olink and mass spectrometry-based cohorts and concluded that large sample sizes are needed to discover more complex genetic effects. Our approach presents a systematic, large-scale attempt to identify complex effects of plasma protein levels and can be applied to study other tissues or molecular QTLs.

show abstract

Section: Introductionmentioning

confidence: 83%

Non-linear genetic regulation of the blood plasma proteome

Sigurdsson,

Gräf,

Yang

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Trait imputation enhances nonlinear genetic prediction for some traits

He,

Fu,

Ren

et al. 2024

Genetics

View full text Add to dashboard Cite

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.

show abstract

Improved prediction of blood biomarkers using deep learning

Cited by 2 publications

References 62 publications

Non-linear genetic regulation of the blood plasma proteome

Non-linear genetic regulation of the blood plasma proteome

Trait imputation enhances nonlinear genetic prediction for some traits

Contact Info

Product

Resources

About