Improving Machine Learning Prediction of ADHD Using Gene Set Polygenic Risk Scores and Risk Scores from Genetically Correlated Phenotypes

Background: Polygenic risk scores (PRSs), which sum the effects of SNPs throughout the genome to measure risk afforded by common genetic variants, have improved our ability to estimate disorder risk for Attention-Deficit/Hyperactivity Disorder (ADHD) but the accuracy of risk prediction is rarely investigated. Methods: With the goal of improving risk prediction, we performed gene set analysis of GWAS data to select gene sets associated with ADHD within a training subset. For each selected gene set, we generated gene set polygenic risk scores (gsPRSs), which sum the effects of SNPs for each selected gene set. We created gsPRS for ADHD and for phenotypes having a high genetic correlation with ADHD. These gsPRS were added to the standard PRS as input to machine learning models predicting ADHD. We used feature importance scores to select gsPRS for a final model and to generate a ranking of the most consistently predictive gsPRS. Results: For a test subset that had not been used for training or validation, a random forest (RF) model using PRSs from ADHD and genetically correlated phenotypes and an optimized group of 20 gsPRS had an area under the receiving operating characteristic curve (AUC) of 0.72 (95% CI: 0.70 to 0.74). This AUC was a statistically significant improvement over logistic regression models and RF models using only PRS from ADHD and genetically correlated phenotypes. Conclusions: Summing risk at the gene set level and incorporating genetic risk from disorders with high genetic correlations with ADHD improved the accuracy of predicting ADHD. Learning curves suggest that additional improvements would be expected with larger study sizes. Our study suggests that better accounting of genetic risk and the genetic context of allelic differences results in more predictive models.

Contributions of common genetic variants to specific languages and to when a language is learned

Research over the past two decades has identified a group of common genetic variants explaining a portion of variance in native language ability. The present study investigates whether the same group of genetic variants are associated with different languages and languages learned at different times in life. We recruited 940 young adults who spoke from childhood Chinese and English as their first (native) (L1) and second (L2) language, respectively, who were learners of a new, third (L3) language. For the variants examined, we found a general decrease of contribution of genes to language functions from native to foreign (L2 and L3) languages, with variance in foreign languages explained largely by non-genetic factors such as musical training and motivation. Furthermore, genetic variants that were found to contribute to traits specific to Chinese and English respectively exerted the strongest effects on L1 and L2. These results seem to speak against the hypothesis of a language- and time-universal genetic core of linguistic functions. Instead, they provide preliminary evidence that genetic contribution to language may depend at least partly on the intricate language-specific features. Future research including a larger sample size, more languages and more genetic variants is required to further explore these hypotheses.

Genetically Based Atrial Fibrillation: Current Considerations for Diagnosis and Management

Atrial fibrillation (AF) is the most common atrial arrhythmia and is subcategorized into numerous clinical phenotypes. Given its heterogeneity, investigations into the genetic mechanisms underlying AF have been pursued in recent decades, with predominant analyses focusing on early onset or lone AF. Linkage analyses, genome wide association studies (GWAS), and single gene analyses have led to the identification of rare and common genetic variants associated with AF risk. Significant overlap with genetic variants implicated in dilated cardiomyopathy syndromes, including truncating variants of the sarcomere protein titin, have been identified through these analyses, in addition to other genes associated with cardiac structure and function. Despite this, widespread utilization of genetic testing in AF remains hindered by the unclear impact of genetic risk identification on clinical outcomes and the high prevalence of variants of unknown significance (VUS). However, genetic testing is a reasonable option for patients with early onset AF and in those with significant family history of arrhythmia. While many knowledge gaps remain, emerging data support genotyping to inform selection of AF therapeutics. In this review we highlight the current understanding of the complex genetic basis of AF and explore the overlap of AF with inherited cardiomyopathy syndromes. We propose a set of criteria for clinical genetic testing in AF patients and outline future steps for the integration of genetics into AF care.

A Saturated Map of Common Genetic Variants Associated with Human Height from 5.4 Million Individuals of Diverse Ancestries

Common SNPs are predicted to collectively explain 40-50% of phenotypic variation in human height, but identifying the specific variants and associated regions requires huge sample sizes. Here we show, using GWAS data from 5.4 million individuals of diverse ancestries, that 12,111 independent SNPs that are significantly associated with height account for nearly all of the common SNP-based heritability. These SNPs are clustered within 7,209 non-overlapping genomic segments with a median size of ~90 kb, covering ~21% of the genome. The density of independent associations varies across the genome and the regions of elevated density are enriched for biologically relevant genes. In out-of-sample estimation and prediction, the 12,111 SNPs account for 40% of phenotypic variance in European ancestry populations but only ~10%-20% in other ancestries. Effect sizes, associated regions, and gene prioritization are similar across ancestries, indicating that reduced prediction accuracy is likely explained by linkage disequilibrium and allele frequency differences within associated regions. Finally, we show that the relevant biological pathways are detectable with smaller sample sizes than needed to implicate causal genes and variants. Overall, this study, the largest GWAS to date, provides an unprecedented saturated map of specific genomic regions containing the vast majority of common height-associated variants.

Common idiopathic pulmonary fibrosis risk variants are associated with hypersensitivity pneumonitis

A subset of patients with hypersensitivity pneumonitis (HP) develop lung fibrosis that is clinically similar to idiopathic pulmonary fibrosis (IPF). To address the aetiological determinants of fibrotic HP, we investigated whether the common IPF genetic risk variants were also relevant in study subjects with fibrotic HP. Our findings indicate that common genetic variants in TERC, DSP, MUC5B and IVD were significantly associated with fibrotic HP. These findings provide support for a shared etiology and pathogenesis between fibrotic HP and IPF.

Stem Cell-Derived β Cells: A Versatile Research Platform to Interrogate the Genetic Basis of β Cell Dysfunction

Pancreatic β cell dysfunction is a central component of diabetes progression. During the last decades, the genetic basis of several monogenic forms of diabetes has been recognized. Genome-wide association studies (GWAS) have also facilitated the identification of common genetic variants associated with an increased risk of diabetes. These studies highlight the importance of impaired β cell function in all forms of diabetes. However, how most of these risk variants confer disease risk, remains unanswered. Understanding the specific contribution of genetic variants and the precise role of their molecular effectors is the next step toward developing treatments that target β cell dysfunction in the era of personalized medicine. Protocols that allow derivation of β cells from pluripotent stem cells, represent a powerful research tool that allows modeling of human development and versatile experimental designs that can be used to shed some light on diabetes pathophysiology. This article reviews different models to study the genetic basis of β cell dysfunction, focusing on the recent advances made possible by stem cell applications in the field of diabetes research.

Analysis of the Batch Effect Due to Sequencing Center in Population Statistics Quantifying Rare Events in the 1000 Genomes Project

The 1000 Genomes Project (1000G) is one of the most popular whole genome sequencing datasets used in different genomics fields and has boosting our knowledge in medical and population genomics, among other fields. Recent studies have reported the presence of ghost mutation signals in the 1000G. Furthermore, studies have shown that these mutations can influence the outcomes of follow-up studies based on the genetic variation of 1000G, such as single nucleotide variants (SNV) imputation. While the overall effect of these ghost mutations can be considered negligible for common genetic variants in many populations, the potential bias remains unclear when studying low frequency genetic variants in the population. In this study, we analyze the effect of the sequencing center in predicted loss of function (LoF) alleles, the number of singletons, and the patterns of archaic introgression in the 1000G. Our results support previous studies showing that the sequencing center is associated with LoF and singletons independent of the population that is considered. Furthermore, we observed that patterns of archaic introgression were distorted for some populations depending on the sequencing center. When analyzing the frequency of SNPs showing extreme patterns of genotype differentiation among centers for CEU, YRI, CHB, and JPT, we observed that the magnitude of the sequencing batch effect was stronger at MAF < 0.2 and showed different profiles between CHB and the other populations. All these results suggest that data from 1000G must be interpreted with caution when considering statistics using variants at low frequency.

Association of the Polygenic Risk Score With the Probability of Prodromal Parkinson’s Disease in Older Adults

Several studies have investigated the association of the Parkinson’s disease (PD) polygenic risk score (PRS) with several aspects of well-established PD. We sought to evaluate the association of PRS with the prodromal stage of PD. We calculated PRS in a longitudinal sample (n = 1120) of community dwelling individuals ≥ 65 years from the HELIAD (The Hellenic Longitudinal Investigation of Aging and Diet) study in order to evaluate the association of this score with the probability of prodromal PD or any of the established risk and prodromal markers in MDS research criteria, using regression multi-adjusted models. Increases in PRS estimated from GWAS summary statistics’ ninety top SNPS with p &lt; 5 × 10–8 was associated with increased odds of having probable/possible prodromal PD (i.e., ≥ 30% probability, OR = 1.033, 95%CI: 1.009–1.057 p = 0.006). From the prodromal PD risk markers, significant association was found between PRS and global cognitive deficit exclusively (p = 0.003). To our knowledge, our study is the first population based study investigating the association between PRS scores and prodromal markers of Parkinson’s disease. Our results suggest a strong relationship between the accumulation of many common genetic variants, as measured by PRS, and cognitive deficits.

The Linkage Phase of the Polymorphism KCNH2-K897T Influences the Electrophysiological Phenotype in hiPSC Models of LQT2

While rare mutations in ion channel genes are primarily responsible for inherited cardiac arrhythmias, common genetic variants are also an important contributor to the clinical heterogeneity observed among mutation carriers. The common single nucleotide polymorphism (SNP) KCNH2-K897T is associated with QT interval duration, but its influence on the disease phenotype in patients with long QT syndrome type 2 (LQT2) remains unclear. Human induced pluripotent stem cells (hiPSCs), coupled with advances in gene editing technologies, are proving an invaluable tool for modeling cardiac genetic diseases and identifying variants responsible for variability in disease expressivity. In this study, we have used isogenic hiPSC-derived cardiomyocytes (hiPSC-CMs) to establish the functional consequences of having the KCNH2-K897T SNP in cis- or trans-orientation with LQT2-causing missense variants either within the pore-loop domain (KCNH2A561T/WT) or tail region (KCNH2N996I/WT) of the potassium ion channel, human ether-a-go-go-related gene (hERG). When KCNH2-K897T was on the same allele (cis) as the primary mutation, the hERG channel in hiPSC-CMs exhibited faster activation and deactivation kinetics compared to their trans-oriented counterparts. Consistent with this, hiPSC-CMs with KCNH2-K897T in cis orientation had longer action and field potential durations. Furthermore, there was an increased occurrence of arrhythmic events upon pharmacological blocking of hERG. Collectively, these results indicate that the common polymorphism KCNH2-K897T differs in its influence on LQT2-causing KCNH2 mutations depending on whether it is present in cis or trans. This study corroborates hiPSC-CMs as a powerful platform to investigate the modifying effects of common genetic variants on inherited cardiac arrhythmias and aids in unraveling their contribution to the variable expressivity of these diseases.