In phenotype prediction the physical characteristics of an organism are predicted from knowledge of its genotype and environment. Such studies, often called genome-wide association studies, are of the highest societal importance, as they are of central importance to medicine, crop-breeding, etc. We investigated three phenotype prediction problems: one simple and clean (yeast), and the other two complex and real-world (rice and wheat). We compared standard machine learning methods; elastic net, ridge regression, lasso regression, random forest, gradient boosting machines (GBM), and support vector machines (SVM), with two state-of-the-art classical statistical genetics methods; genomic BLUP and a two-step sequential method based on linear regression. Additionally, using the clean yeast data, we investigated how performance varied with the complexity of the biological mechanism, the amount of observational noise, the number of examples, the amount of missing data, and the use of different data representations. We found that for almost all the phenotypes considered, standard machine learning methods outperformed the methods from classical statistical genetics. On the yeast problem, the most successful method was GBM, followed by lasso regression, and the two statistical genetics methods; with greater mechanistic complexity GBM was best, while in simpler cases lasso was superior. In the wheat and rice studies the best two methods were SVM and BLUP. The most robust method in the presence of noise, missing data, etc. was random forests. The classical statistical genetics method of genomic BLUP was found to perform well on problems where there was population structure. This suggests that standard machine learning methods need to be refined to include population structure information when this is present. We conclude that the application of machine learning methods to phenotype prediction problems holds great promise, but that determining which methods is likely to perform well on any given problem is elusive and non-trivial.

Perennial ryegrass (Lolium perenne L.) is one of the most widely grown forage grasses in temperate agriculture. In order to maintain and increase its usage as forage in livestock agriculture, there is a continued need for improvement in biomass yield, quality, disease resistance and seed yield. Genetic gain for traits such as biomass yield has been relatively modest. This has been attributed to its long breeding cycle, and the necessity to use population based breeding methods. Thanks to recent advances in genotyping techniques there is increasing interest in genomic selection from which genomically estimated breeding values (GEBV) are derived. In this paper we compare the classical RRBLUP model with state-of-the-art machine learning (ML) techniques that should yield themselves easily to use in GS and demonstrate their application to predicting quantitative traits in a breeding population of L. perenne. Prediction accuracies varied from 0 to 0.59 depending on trait, prediction model and composition of the training population. The BLUP model produced the highest prediction accuracies for most traits and training populations. Forage quality traits had the highest accuracies compared to yield related traits. There appeared to be no clear pattern to the effect of the training population composition on the prediction accuracies. The heritability of the forage quality traits was generally higher than for the yield related traits, and could partly explain the difference in accuracy. Some population structure was evident in the breeding populations, and probably contributed to the varying effects of training population on the predictions. The average linkage disequilibrium (LD) between adjacent markers ranged from 0.121 to 0.215. Higher marker density and larger training population closely related with the test population are likely to improve the prediction accuracy.

Background: In Sweden, social restrictions to contain SARS‐CoV‐2 have primarily relied upon voluntary adherence to a set of recommendations. Strict lockdowns have not been enforced, potentially affecting viral dissemination. To understand the levels of past SARS‐CoV‐2 infection in the Stockholm population before the start of mass vaccinations, healthy blood donors and pregnant women (n = 5,100) were sampled at random between 14 March 2020 and 28 February 2021.Methods: In this cross‐sectional prospective study, otherwise‐healthy blood donors (n = 2,600) and pregnant women (n = 2,500) were sampled for consecutive weeks (at four intervals) throughout the study period. Sera from all participants and a cohort of historical (negative) controls (n = 595) were screened for IgG responses against stabilized trimers of the SARS‐CoV‐2 spike (S) glycoprotein and the smaller receptor‐binding domain (RBD). As a complement to standard analytical approaches, a probabilistic (cut‐off independent) Bayesian framework that assigns likelihood of past infection was used to analyse data over time.Setting: Healthy participant samples were randomly selected from their respective pools through Karolinska University Hospital. The study was carried out in accordance with Swedish Ethical Review Authority: registration number 2020–01807.Participants: No participants were symptomatic at sampling, and blood donors were all over the age of 18. No additional metadata were available from the participants.Results: Blood donors and pregnant women showed a similar seroprevalence. After a steep rise at the start of the pandemic, the seroprevalence trajectory increased steadily in approach to the winter second wave of infections, approaching 15% of all individuals surveyed by 13 December 2020. By the end of February 2021, 19% of the population tested seropositive. Notably, 96% of seropositive healthy donors screened (n = 56) developed neutralizing antibody responses at titres comparable to or higher than those observed in clinical trials of SARS‐CoV‐2 spike mRNA vaccination, supporting that mild infection engenders a competent B‐cell response.Conclusions: These data indicate that in the first year since the start of community transmission, seropositivity levels in metropolitan in Stockholm had reached approximately one in five persons, providing important baseline seroprevalence information prior to the start of vaccination.

Thousands of genetic variants are associated with human disease risk, but linkage disequilibrium (LD) hinders fine-mapping the causal variants. Both lack of power, and joint tagging of two or more distinct causal variants by a single non-causal SNP, lead to inaccuracies in fine-mapping, with stochastic search more robust than stepwise. We develop a computationally efficient multinomial fine-mapping (MFM) approach that borrows information between diseases in a Bayesian framework. We show that MFM has greater accuracy than single disease analysis when shared causal variants exist, and negligible loss of precision otherwise. MFM analysis of six immune-mediated diseases reveals causal variants undetected in individual disease analysis, including in IL2RA where we confirm functional effects of multiple causal variants using allele-specific expression in sorted CD4⁺ T cells from genotype-selected individuals. MFM has the potential to increase fine-mapping resolution in related diseases enabling the identification of associated cellular and molecular phenotypes.

Identifying DNA cis-regulatory modules (CRMs) that control the expression of specific genes is crucial for deciphering the logic of transcriptional control. Natural genetic variation can point to the possible gene regulatory function of specific sequences through their allelic associations with gene expression. However, comprehensive identification of causal regulatory sequences in brute-force association testing without incorporating prior knowledge is challenging due to limited statistical power and effects of linkage disequilibrium. Sequence variants affecting transcription factor (TF) binding at CRMs have a strong potential to influence gene regulatory function, which provides a motivation for prioritizing such variants in association testing. Here, we generate an atlas of CRMs showing predicted allelic variation in TF binding affinity in human lymphoblastoid cell lines and test their association with the expression of their putative target genes inferred from Promoter Capture Hi-C and immediate linear proximity. We reveal >1300 CRM TF-binding variants associated with target gene expression, the majority of them undetected with standard association testing. A large proportion of CRMs showing associations with the expression of genes they contact in 3D localize to the promoter regions of other genes, supporting the notion of ‘epromoters’: dual-action CRMs with promoter and distal enhancer activity.

A transcriptome‐wide association study (TWAS) attempts to identify disease associated genes by imputing gene expression into a genome‐wide association study (GWAS) using an expression quantitative trait loci (eQTL) data set and then testing for associations with a trait of interest. Regulatory processes may be shared across related tissues and one natural extension of TWAS is harnessing cross‐tissue correlation in gene expression to improve prediction accuracy. Here, we studied multi‐tissue extensions of lasso regression and random forests (RF), joint lasso and RF‐MTL (multi‐task learning RF), respectively. We found that, on our chosen eQTL data set, multi‐tissue methods were generally more accurate than their single‐tissue counterparts, with RF‐MTL performing the best. Simulations showed that these benefits generally translated into more associated genes identified, although highlighted that joint lasso had a tendency to erroneously identify genes in one tissue if there existed an eQTL signal for that gene in another. Applying the four methods to a type 1 diabetes GWAS, we found that multi‐tissue methods found more unique associated genes for most of the tissues considered. We conclude that multi‐tissue methods are competitive and, for some cell types, superior to single‐tissue approaches and hold much promise for TWAS studies.