Ascertainment Bias Research Articles

KIN: a method to infer relatedness from low-coverage ancient DNA

Genetic kinship of ancient individuals can provide insights into their culture and social hierarchy, and is relevant for downstream genetic analyses. However, estimating relatedness from ancient DNA is difficult due to low-coverage, ascertainment bias, or contamination from various sources. Here, we present KIN, a method to estimate the relatedness of a pair of individuals from the identical-by-descent segments they share. KIN accurately classifies up to 3rd-degree relatives using at least 0.05x sequence coverage and differentiates siblings from parent-child pairs. It incorporates additional models to adjust for contamination and detect inbreeding, which improves classification accuracy.

COVID-19 vaccination is effective at preventing severe illness and complications during pregnancy

COVID-19 vaccination is effective at preventing severe illness and complications during pregnancy

Constraining the galaxy-halo connection of infrared-selected unWISE galaxies with galaxy clustering and galaxy-CMB lensing power spectra

We present the first detailed analysis of the connection between galaxies and their dark matter halos for the unWISE galaxy catalog -- a full-sky, infrared-selected sample built from WISE data, containing over 500 million galaxies. Using unWISE galaxy-galaxy auto-correlation and Planck CMB lensing-galaxy cross-correlation measurements down to 10 arcmin angular scales, we constrain the halo occupation distribution (HOD), a model describing how central and satellite galaxies are distributed within dark matter halos, for three unWISE} galaxy samples at mean redshifts $\bar{z} \approx 0.6$, $1.1$, and $1.5$. We constrain the characteristic minimum halo mass to host a central galaxy, $M_\mathrm{min}^\mathrm{HOD} = 1.83^{+0.41}_{-1.63} \times 10^{12} M_\odot/h$, $5.22^{+0.34}_{-4.80} \times 10^{12} M_\odot/h$, $6.60 ^{+0.30}_{-1.11} \times 10^{13} M_\odot/h$ for the unWISE samples at $\bar{z}\approx 0.6$, $1.1$, and $1.5$, respectively. We find that all three samples are dominated by central galaxies, rather than satellites. Using our constrained HOD models, we infer the effective linear galaxy bias for each unWISE sample, and find that it does not evolve as steeply with redshift as found in previous perturbation-theory-based analyses of these galaxies. We discuss possible sources of systematic uncertainty in our results, the most significant of which is the uncertainty on the galaxy redshift distribution. Our HOD constraints provide a detailed, quantitative understanding of how the unWISE galaxies populate the underlying dark matter halo distribution. These constraints will have a direct impact on future studies employing the unWISE galaxies as a cosmological and astrophysical probe, including measurements of ionized gas thermodynamics and dark matter profiles via Sunyaev-Zel'dovich and lensing cross-correlations.

Calculating variant penetrance from family history of disease and average family size in population-scale data

BackgroundGenetic penetrance is the probability of a phenotype when harbouring a particular pathogenic variant. Accurate penetrance estimates are important across biomedical fields including genetic counselling, disease research, and gene therapy. However, existing approaches for penetrance estimation require, for instance, large family pedigrees or availability of large databases of people affected and not affected by a disease.MethodsWe present a method for penetrance estimation in autosomal dominant phenotypes. It examines the distribution of a variant among people affected (cases) and unaffected (controls) by a phenotype within population-scale data and can be operated using cases only by considering family disease history. It is validated through simulation studies and candidate variant-disease case studies.ResultsOur method yields penetrance estimates which align with those obtained via existing approaches in the Parkinson’s disease LRRK2 gene and pulmonary arterial hypertension BMPR2 gene case studies. In the amyotrophic lateral sclerosis case studies, examining penetrance for variants in the SOD1 and C9orf72 genes, we make novel penetrance estimates which correspond closely to understanding of the disease.ConclusionsThe present approach broadens the spectrum of traits for which reliable penetrance estimates can be obtained. It has substantial utility for facilitating the characterisation of disease risks associated with rare variants with an autosomal dominant inheritance pattern. The yielded estimates avoid any kinship-specific effects and can circumvent ascertainment biases common when sampling rare variants among control populations.

Cross‐ancestry effect of APOE on cognitive performance

AbstractBackgroundAlzheimer's disease (AD) is more prevalent in populations of non‐European ancestry, with Black and Hispanic adults 1.5 and 2.5 times more likely to develop AD respectively. Health disparities in AD could be explained by race and ancestry, where race captures risk factors due to socio‐economic aspects of health and ancestry captures the biological mechanisms influenced by genetic variation. The APOE ε4 allele is the most significant genetic risk factor for AD; however, the risk conferred by APOE ε4 differs across populations. Here, we investigate the interactive effect of race, ancestry and APOE on episodic memory performance.MethodsEpisodic memory at last assessment was measured using the Consortium to Establish a Registry for Alzheimer’s Disease Word List Memory Task in ∼500 participants from the Alzheimer's Disease Research Center (ADRC) at the Icahn School of Medicine at Mount Sinai (ISMMS), including 82 African American, 118 Hispanics, 117 Asian Americans, and 275 White participants. Genetic ancestry was determined using principal component analysis; global ancestry proportions using ADMIXTURE; and local ancestry in the APOE locus using RFMix. The interactive effect of APOE and genetic ancestry, global ancestry proportions, and local ancestry locus on episodic memory performance was evaluated using linear regression models adjusting for age, sex, education, telehealth/in‐person visit, and year of visit.ResultsAdmixed American and African genetic ancestry assignments showed an interaction with APOE (African: p‐value=0.023 and Amerindian: p‐value=0.002). In addition, higher proportion of African and Admixed American global ancestry showed significant interactions with APOE (African: p‐value=0.032 and Admixed American: p‐value=0.029). Lastly, ancestry local to APOE region showed an interaction with the APOE e4 allele in Admixed Americans such that an Admixed American background was associated with better cognitive performance in comparison to a European ancestral background (p‐value=0.008).ConclusionsOur study showed that ancestry modifies the effect of APOE on cognitive performance. Future work needs to evaluate potential confounders associated with social determinants of health and ascertainment bias during enrollment, including age at assessment and visit type. In sum, this study corroborates a role of ancestry in modifying APOE associated AD risk.

Neonatal exposures to sevoflurane in rhesus monkeys alter synaptic ultrastructure in later life.

Repeated or prolonged early life exposure to anesthesia is neurotoxic in animals and associated with neurocognitive impairment in later life in humans. We used electron microscopy with unbiased stereological sampling to assess synaptic ultrastructure in dorsolateral prefrontal cortex (dlPFC) and hippocampal CA1 of female and male rhesus monkeys, four years after three 4-h exposures to sevoflurane during the first five postnatal weeks. This allowed us to ascertain long-term consequences of anesthesia exposure without confounding effects of surgery or illness. Synapse areas were reduced in the largest synapses in CA1 and dlPFC, predominantly in perforated spinous synapses in CA1 and nonperforated spinous synapses in dlPFC. Mitochondrial morphology and localization changed subtly in both areas. Synapse areas in CA1 correlated with response to a mild social stressor. Thus, exposure to anesthesia in infancy can cause long-term ultrastructural changes in primates, which may be substrates for long-term alterations in synaptic transmission and behavioral deficits.

It Is Time to Rethink Adult Glaucoma Screening Recommendations.

It Is Time to Rethink Adult Glaucoma Screening Recommendations.

Phenotypic and genetic factors associated with donation of DNA and consent to record linkage for prescription history in the Australian Genetics of Depression Study.

Samples can be prone to ascertainment and attrition biases. The Australian Genetics of Depression Study is a large publicly recruited cohort (n = 20,689) established to increase the understanding of depression and antidepressant treatment response. This study investigates differences between participants who donated a saliva sample or agreed to linkage of their records compared to those who did not. We observed that older, male participants with higher education were more likely to donate a saliva sample. Self-reported bipolar disorder, ADHD, panic disorder, PTSD, substance use disorder, and social anxiety disorder were associated with lower odds of donating a saliva sample, whereas anorexia was associated with higher odds of donation. Male and younger participants showed higher odds of agreeing to record linkage. Participants with higher neuroticism scores and those with a history of bipolar disorder were also more likely to agree to record linkage whereas participants with a diagnosis of anorexia were less likely to agree. Increased likelihood of consent was associated with increased genetic susceptibility to anorexia and reduced genetic risk for depression, and schizophrenia. Overall, our results show moderate differences among these subsamples. Most current epidemiological studies do not search for attrition biases at the genetic level. The possibility to do so is a strength of samples such as the AGDS. Our results suggest that analyses can be made more robust by identifying attrition biases both on the phenotypic and genetic level, and either contextualising them as a potential limitation or performing sensitivity analyses adjusting for them.

The Influence of Internet Usage on Gender Division of Labour Bias: Evidence from Chinese General Social Survey Data

Based on data from 18,741 respondents in the 2010 and 2017 Chinese General Social Survey, this paper empirically analyzes the influence of Internet usage on gender division of labour bias by using the propensity score matching model and tests the sensitivity of the estimated results. The results show that firstly, a series of personal characteristics such as gender, age, years of education, ethnicity, political status, health status, marital status, household registration status, labour income, mother’s years of education, and family economic status will have a significant impact on whether they frequently use the Internet. Secondly, frequent internet usage can significantly reduce respondents’ agreement of gender division of labour bias, indicating that frequent internet usage can significantly improve respondents’ gender division of labour bias, and the results of Rosenbaum’s bound test and H-L confidence interval test both support the above conclusions. Thirdly, frequent internet usage can significantly reduce the recognition degree of women respondents and men respondents to gender division of labour bias, which indicates that frequent internet usage can not only strengthen women’s self-identity to a great extent but also improve men’s exclusion bias to a certain extent. Therefore, this paper puts forward some suggestions from two aspects: improving the accessibility of internet usage and guiding and controlling the content and behavior of various online media.

Current global population size, post-whaling trend and historical trajectory of sperm whales

The sperm whale lives in most deep ice-free waters of the globe. It was targeted during two periods of whaling peaking in the 1840’s and 1960’s. Using a habitat suitability model, we extrapolated estimates of abundance from visual and acoustic surveys to give a global estimate of 736,053 sperm whales (CV = 0.218) in 1993. Estimates of trends in the post-whaling era suggest that: whaling, by affecting the sex ratio and/or the social cohesion of females, reduced recovery rates well after whaling ceased; preferentially-targeted adult males show the best evidence of recovery, presumably due to recruitment from breeding populations; several decades post-whaling, sperm whale populations not facing much human impact are recovering slowly, but populations may be declining in areas with substantial anthropogenic footprint. A theta-logistic population model enhanced to simulate spatial structure and the non-removal impacts of whaling indicated a pre-whaling population of 1,949,698 (CV = 0.178) in 1710 being reduced by whaling, and then then recovering a little to about 844,761 (CV = 0.209) in 2022. There is much uncertainty about these numbers and trends. A larger population estimate than produced by a similar analysis in 2002 is principally due to a better assessment of ascertainment bias.

Developing a reference standard for pertussis by applying a stratified sampling strategy to electronic medical record data

Pertussis surveillance remains essential in Canada, but ascertainment bias limits the accuracy of surveillance data. Introducing other sources to improve detection has highlighted the importance of validation. However, challenges arise due to low prevalence, and oversampling suspected cases can introduce partial verification bias. The aim of this study was to build a reference standard for pertussis validation studies that provides adequate analytic precision and minimizes bias. We used a stratified strategy to sample the reference standard from a primary care electronic medical record cohort. We incorporated abstractor notes into definite, possible, ruled-out, and no mention of pertussis classifications which were based on surveillance case definitions. We abstracted eight hundred records from the cohort of 404,922. There were 208 (26%) definite and 261 (32.6%) possible prevalent pertussis cases. Classifications demonstrated a wide variety of case severities. Abstraction reliability was moderate to substantial based on Cohen's kappa and raw percent agreement. When conducting validation studies for pertussis and other low prevalence diseases, this stratified sampling strategy can be used to develop a reference standard using limited resources. This approach mitigates verification and spectrum bias while providing sufficient precision and incorporating a range of case severities.

Prenatal diagnosis of sex chromosome aneuploidy-What do we tell the prospective parents?

Sex chromosome aneuploidy (SCA) can be detected on prenatal diagnostic testing and cell free DNA screening (cfDNA). High risk cfDNA results should be confirmed with diagnostic testing. This summary article serves as an update for prenatal providers and assimilates data from neurodevelopmental, epidemiologic, and registry studies on the most common SCA. This information can be helpful for counseling after prenatal diagnosis of sex chromosome aneuploidy. Incidence estimates may be influenced by ascertainment bias and this article is not a substitute for interdisciplinary consultation and counseling.

Federated learning based on stratified sampling and regularization

Federated learning (FL) is a new distributed learning framework that is different from traditional distributed machine learning: (1) differences in communication, computing, and storage performance among devices (device heterogeneity), (2) differences in data distribution and data volume (data heterogeneity), and (3) high communication consumption. Under heterogeneous conditions, the data distribution of clients varies greatly, which leads to the problem that the convergence speed of the training model decreases and the training model cannot converge to the global optimal solution. In this work, an FL algorithm based on stratified sampling and regularization (FedSSAR) is proposed. In FedSSAR, a density-based clustering method is used to divide the overall client into different clusters, then, some available clients are proportionally extracted from different clusters to participate in training which realizes unbiased sampling for the overall client and reduces the aggregation weight variance of the client. At the same time, when calculating the model local loss function, we limit the update direction of the model by a regular term, so that heterogeneous clients are optimized in the globally optimal direction. We prove the convergence of FedSSAR theoretically and experimentally, and demonstrate the superiority of FedSSAR by comparing it with other FL algorithms on public datasets.

Extensively Explored and Evaluated Actor-Critic With Expert-Guided Policy Learning and Fuzzy Feedback Reward for Robotic Trajectory Generation

Trajectory generation for redundant manipulators based on inverse kinematics (IK) still faces some restraints, as it lacks universal IK calculation or specific trajectory generation methods that are suitable for robots with arbitrary degrees of freedom. In this article, the IK-free trajectory generation for robot manipulators is formulated as a Markov decision process and implemented by a general method based on deep reinforcement learning. First, an extensively explored and evaluated actor-critic (E3AC) algorithm that can make diverse action explorations and comprehensive evaluations is designed to solve the trajectory generation problem. Second, a dual-memory structure with expert-guided policy learning strategy is proposed to further enhance the performance of the algorithm in the early training period by additional successful experiences and performing an increasingly unbiased data sampling. Third, a fuzzy feedback reward mechanism that can directly establish a mapping from the abundant state variables to the self-tuning reward is constructed, instead of puzzling out an explicit function to feature the complex relations among the control objects. Finally, the comparative experimental results show that, the proposed approach is more efficient in algorithm convergence and reward calculation, and is more qualified for complex tasks with strong randomness.

Predicting Real-world Severe Hypoglycemia Risk in Diabetes (iNPHORM, USA)

In the United States (US), prognostic models for iatrogenic severe hypoglycemia (SH) derive from secondary sources subject to poor generalizability, data gaps and outcome ascertainment bias. Redressing this gap, we conducted the primary, longitudinal iNPHORM study.

Klinefelter Syndrome: What should we tell prospective parents?

Klinefelter syndrome (KS) or 47,XXY is the most common sex chromosome aneuploidy (SCA), occurring at a prevalence of 1 in 600 male pregnancies. Historically, only 25% of individuals with KS came to medical attention, for a range of issues across the life course including under-virilisation at birth, developmental and social concerns in childhood, absence, delay or arrest of puberty in adolescence or infertility in adulthood. Our understanding of the phenotypic spectrum of KS has been largely influenced by this ascertainment bias. With increasing uptake of antenatal noninvasive prenatal testing (NIPT), a corresponding increase in identification of KS has been documented. Population-based longitudinal data from infancy to adulthood on these individuals is lacking, which impedes balanced antenatal genetic counselling and raises issues for prospective parents and clinicians alike.

Reduced penetrance of MODY-associated HNF1A/HNF4A variants but not GCK variants in clinically unselected cohorts

The true prevalence and penetrance of monogenic disease variants are often not known because of clinical-referral ascertainment bias. We comprehensively assess the penetrance and prevalence of pathogenic variants in HNF1A, HNF4A, and GCK that account for >80% of monogenic diabetes. We analyzed clinical and genetic data from 1,742 clinically referred probands, 2,194 family members, clinically unselected individuals from a US health system-based cohort (n= 132,194), and a UK population-based cohort (n= 198,748). We show that one in 1,500 individuals harbor a pathogenic variant in one of these genes. The penetrance of diabetes for HNF1A and HNF4A pathogenic variants was substantially lower in the clinically unselected individuals compared to clinically referred probands and was dependent on the setting (32% in the population, 49% in the health system cohort, 86% in a family member, and 98% in probands for HNF1A). The relative risk of diabetes was similar across the clinically unselected cohorts highlighting the role of environment/other genetic factors. Surprisingly, the penetrance of pathogenic GCK variants was similar across all cohorts (89%-97%). We highlight that pathogenic variants in HNF1A, HNF4A, and GCK are not ultra-rare in the population. For HNF1A and HNF4A, we need to tailor genetic interpretation and counseling based on the setting in which a pathogenic monogenic variant was identified. GCK is an exception with near-complete penetrance in all settings. This along with the clinical implication of diagnosis makes it an excellent candidate for the American College of Medical Genetics secondary gene list.

Quantum Annealing Sampling with a Bias Field

The presence of a bias field, encoding some information about the target state, can enhance the performance of quantum optimization methods. Here we investigate the effect of such a bias field on the outcome of quantum annealing sampling, at the example of the exact cover problem. The sampling is carried out on a D-Wave machine, and different bias configurations are benchmarked against the unbiased sampling procedure. It is found that the biased annealing algorithm works particularly well for larger problem sizes, where the Hamming distance between bias and target configuration becomes less important. This work motivates future research efforts for finding good bias configurations, either on the quantum machine itself, or in a hybrid fashion via classical algorithms.

A Comprehensive Assessment of Qualitative and Quantitative Prodromal Parkinsonian Features in Carriers of Gaucher Disease-Identifying Those at the Greatest Risk.

Carriers of GBA1 gene variants have a significant risk of developing Parkinson’s disease (PD). A cohort study of GBA carriers between 40–75 years of age was initiated to study the presence of prodromal PD features. Participants underwent non-invasive tests to assess different domains of PD. Ninety-eight unrelated GBA carriers were enrolled (43 males) at a median age (range) of 51 (40–74) years; 71 carried the N370S variant (c.1226A > G) and 25 had a positive family history of PD. The Montreal Cognitive Assessment (MoCA) was the most frequently abnormal (23.7%, 95% CI 15.7–33.4%), followed by the ultrasound hyperechogenicity (22%, 95% CI 14–32%), Unified Parkinson’s Disease Rating Scale part III (UPDRS-III) (17.2%, 95% CI 10.2–26.4%), smell assessment (12.4%, 95% CI 6.6–20.6%) and abnormalities in sleep questionnaires (11%, 95% CI 5.7–19.4%). Significant correlations were found between tests from different domains. To define the risk for PD, we assessed the bottom 10th percentile of each prodromal test, defining this level as “abnormal”. Then we calculated the percentage of “abnormal” tests for each subject; the median (range) was 4.55 (0–43.5%). Twenty-two subjects had more than 15% “abnormal” tests. The limitations of the study included ascertainment bias of individuals with GBA-related PD in relatives, some incomplete data due to technical issues, and a lack of well-characterized normal value ranges in some tests. We plan to enroll additional participants and conduct longitudinal follow-up assessments to build a model for identifying individuals at risk for PD and investigate interventions aiming to delay the onset or perhaps to prevent full-blown PD.

Viral Coinfections in Hospitalized Coronavirus Disease 2019 Patients Recruited to the International Severe Acute Respiratory and Emerging Infections Consortium WHO Clinical Characterisation Protocol UK Study.

We conducted this study to assess the prevalence of viral coinfection in a well characterized cohort of hospitalized coronavirus disease 2019 (COVID-19) patients and to investigate the impact of coinfection on disease severity. Multiplex real-time polymerase chain reaction testing for endemic respiratory viruses was performed on upper respiratory tract samples from 1002 patients with COVID-19, aged <1 year to 102 years old, recruited to the International Severe Acute Respiratory and Emerging Infections Consortium WHO Clinical Characterisation Protocol UK study. Comprehensive demographic, clinical, and outcome data were collected prospectively up to 28 days post discharge. A coinfecting virus was detected in 20 (2.0%) participants. Multivariable analysis revealed no significant risk factors for coinfection, although this may be due to rarity of coinfection. Likewise, ordinal logistic regression analysis did not demonstrate a significant association between coinfection and increased disease severity. Viral coinfection was rare among hospitalized COVID-19 patients in the United Kingdom during the first 18 months of the pandemic. With unbiased prospective sampling, we found no evidence of an association between viral coinfection and disease severity. Public health interventions disrupted normal seasonal transmission of respiratory viruses; relaxation of these measures mean it will be important to monitor the prevalence and impact of respiratory viral coinfections going forward.