Upward bias in estimation of genetic effects.
Ontology highlight
ABSTRACT:
SUBMITTER: Siegmund D
PROVIDER: S-EPMC385094 | biostudies-other | 2002 Nov
REPOSITORIES: biostudies-other
Similar Datasets
Project description:BackgroundRandomized trials are considered the gold standard for making inferences about the causal effects of treatments. However, when protocol deviations occur, the baseline randomization of the trial is no longer sufficient to ensure unbiased estimation of the per-protocol effect: post-randomization, time-varying confounders must be sufficiently measured and adjusted for in the analysis. Given the historical emphasis on intention-to-treat effects in randomized trials, measurement of post-randomization confounders is typically infrequent. This may induce bias in estimates of the per-protocol effect, even using methods such as inverse probability weighting, which appropriately account for time-varying confounders affected by past treatment.Methods/designIn order to concretely illustrate the potential magnitude of bias due to infrequent measurement of time-varying covariates, we simulated data from a very large trial with a survival outcome and time-varying confounding affected by past treatment. We generated the data such that the true underlying per-protocol effect is null and under varying degrees of confounding (strong, moderate, weak). In the simulated data, we estimated per-protocol survival curves and associated contrasts using inverse probability weighting under monthly measurement of the time-varying covariates (which constituted complete measurement in our simulation), yearly measurement, as well as 3- and 6-month intervals.ResultsUsing inverse probability weighting, we were able to recover the true null under the complete measurement scenario no matter the strength of confounding. Under yearly measurement intervals, the estimate of the per-protocol effect diverged from the null; inverse probability weighted estimates of the per-protocol 5-year risk ratio based on yearly measurement were 1.19, 1.12, and 1.03 under strong, moderate, and weak confounding, respectively. Bias decreased with measurement interval length. Under all scenarios, inverse probability weighted estimators were considerably less biased than a naive estimator that ignored time-varying confounding completely.ConclusionsBias that arises from interval measurement designs highlights the need for planning in the design of randomized trials for collection of time-varying covariate data. This may come from more frequent in-person measurement or external sources (e.g., electronic medical record data). Such planning will provide improved estimates of the per-protocol effect through the use of methods that appropriately adjust for time-varying confounders.
| S-EPMC6729016 | biostudies-literature
Project description:PurposeThe purpose of this work was to explore the impact of slice profile effects on apparent diffusion coefficient (ADC) mapping of hyperpolarized (HP) substrates.MethodsSlice profile effects were simulated using a Gaussian radiofrequency (RF) pulse with a variety of flip angle schedules and b-value ordering schemes. A long T1 water phantom was used to validate the simulation results, and ADC mapping of HP [13 C,15 N2 ]urea was performed on the murine liver to assess these effects in vivo.ResultsSlice profile effects result in excess signal after repeated RF pulses, causing bias in HP measurements. The largest error occurs for metabolites with small ADCs, resulting in up to 10-fold overestimation for metabolites that are in more-restricted environments. A mixed b-value scheme substantially reduces this bias, whereas scaling the slice-select gradient can mitigate it completely. In vivo, the liver ADC of hyperpolarized [13 C,15 N2 ]urea is nearly 70% lower (0.99 ± 0.22 vs 1.69 ± 0.21 × 10-3 mm2 /s) when slice-select gradient scaling is used.ConclusionSlice profile effects can lead to bias in HP ADC measurements. A mixed b-value ordering scheme can reduce this bias compared to sequential b-value ordering. Slice-select gradient scaling can also correct for this deviation, minimizing bias and providing more-precise ADC measurements of HP substrates. Magn Reson Med 78:1087-1092, 2017. © 2016 International Society for Magnetic Resonance in Medicine.
| S-EPMC5391314 | biostudies-literature
Project description:I analyzed various site pattern combinations in a 4-OTU case to identify sources of starless bias and parameter-estimation bias in likelihood-based phylogenetic methods, and reported three significant contributions. First, the likelihood method is counterintuitive in that it may not generate a star tree with sequences that are equidistant from each other. This behaviour, dubbed starless bias, happens in a 4-OTU tree when there is an excess (i.e., more than expected from a star tree and a substitution model) of conflicting phylogenetic signals supporting the three resolved topologies equally. Special site pattern combinations leading to rejection of a star tree, when sequences are equidistant from each other, were identified. Second, fitting gamma distribution to model rate heterogeneity over sites is strongly confounded with tree topology, especially in conjunction with the starless bias. I present examples to show dramatic differences in the estimated shape parameter α between a star tree and a resolved tree. There may be no rate heterogeneity over sites (with the estimated α > 10000) when a star tree is imposed, but α < 1 (suggesting strong rate heterogeneity over sites) when an (incorrect) resolved tree is imposed. Thus, the dependence of "rate heterogeneity" on tree topology implies that "rate heterogeneity" is not a sequence-specific feature, cautioning against interpreting a small α to mean that some sites are under strong purifying selection and others not. Thirdly, because there is no existing (and working) likelihood method for evaluating a star tree with continuous gamma-distributed rate, I have implemented the method for JC69 in a self-contained R script for a four-OTU tree (star or resolved), in addition to another R script assuming a constant rate over sites. These R scripts should be useful for teaching and exploring likelihood methods in phylogenetics.
| S-EPMC6690233 | biostudies-literature
Project description:MotivationHigh-throughput reporter assays dramatically improve our ability to assign function to noncoding genetic variants, by measuring allelic effects on gene expression in the controlled setting of a reporter gene. Unlike genetic association tests, such assays are not confounded by linkage disequilibrium when loci are independently assayed. These methods can thus improve the identification of causal disease mutations. While work continues on improving experimental aspects of these assays, less effort has gone into developing methods for assessing the statistical significance of assay results, particularly in the case of rare variants captured from patient DNA.ResultsWe describe a Bayesian hierarchical model, called Bayesian Inference of Regulatory Differences, which integrates prior information and explicitly accounts for variability between experimental replicates. The model produces substantially more accurate predictions than existing methods when allele frequencies are low, which is of clear advantage in the search for disease-causing variants in DNA captured from patient cohorts. Using the model, we demonstrate a clear tradeoff between variant sequencing coverage and numbers of biological replicates, and we show that the use of additional biological replicates decreases variance in estimates of effect size, due to the properties of the Poisson-binomial distribution. We also provide a power and sample size calculator, which facilitates decision making in experimental design parameters.Availability and implementationThe software is freely available from www.geneprediction.org/bird. The experimental design web tool can be accessed at http://67.159.92.22:8080.Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC7999138 | biostudies-literature
Project description:Observational case-control studies for biomarker discovery in cancer studies often collect data that are sampled separately from the case and control populations. We present an analysis of the bias in the estimation of the precision of classifiers designed on separately sampled data. The analysis consists of both theoretical and numerical results, which show that classifier precision estimates can display strong bias under separating sampling, with the bias magnitude depending on the difference between the true case prevalence in the population and the sample prevalence in the data. We show that this bias is systematic in the sense that it cannot be reduced by increasing sample size. If information about the true case prevalence is available from public health records, then a modified precision estimator that uses the known prevalence displays smaller bias, which can in fact be reduced to zero as sample size increases under regularity conditions on the classification algorithm. The accuracy of the theoretical analysis and the performance of the precision estimators under separate sampling are confirmed by numerical experiments using synthetic and real data from published observational case-control studies. The results with real data confirmed that under separately sampled data, the usual estimator produces larger, ie, more optimistic, precision estimates than the estimator using the true prevalence value.
| S-EPMC6636226 | biostudies-literature
Project description:Accurate age estimates underpin our everyday social interactions, the provision of age-restricted services and police investigations. Previous work suggests that these judgements are error-prone, but the processes giving rise to these errors are not understood. Here, we present the first systematic test of bias in age estimation using a large database of standardized passport images of heterogeneous ages (n = 3948). In three experiments, we tested a range of perceiver age groups (n = 84), and found average age estimation error to be approximately 8 years. We show that this error can be attributed to two separable sources of bias. First, and accounting for the vast majority of variance, our results show an assimilative serial dependency whereby estimates are systematically biased towards the age of the preceding face. Second, younger faces are generally perceived to be older than they are, and older faces to be younger. In combination, these biases account for around 95% of variance in age estimates. We conclude that perception of age is modulated by representations that encode both a viewer's recent and normative exposure to faces. The finding that age perception is subject to strong top-down influences based on our immediate experience has implications for our understanding of perceptual processes involved in face perception, and for improving accuracy of age estimation in important real-world tasks.
| S-EPMC6227935 | biostudies-literature
Project description:PurposeTo develop neural network (NN)-based quantitative MRI parameter estimators with minimal bias and a variance close to the Cramér-Rao bound.Theory and methodsWe generalize the mean squared error loss to control the bias and variance of the NN's estimates, which involves averaging over multiple noise realizations of the same measurements during training. Bias and variance properties of the resulting NNs are studied for two neuroimaging applications.ResultsIn simulations, the proposed strategy reduces the estimates' bias throughout parameter space and achieves a variance close to the Cramér-Rao bound. In vivo, we observe good concordance between parameter maps estimated with the proposed NNs and traditional estimators, such as non-linear least-squares fitting, while state-of-the-art NNs show larger deviations.ConclusionThe proposed NNs have greatly reduced bias compared to those trained using the mean squared error and offer significantly improved computational efficiency over traditional estimators with comparable or better accuracy.
| S-EPMC10925387 | biostudies-literature
Project description:Studies of recurrent or subsequent disease events may be susceptible to bias caused by selection of subjects who both experience and survive the primary indexing event. Currently, the magnitude of any selection bias, particularly for subsequent time-to-event analysis in genetic association studies, is unknown. We used empirically inspired simulation studies to explore the impact of selection bias on the marginal hazard ratio for risk of subsequent events among those with established coronary heart disease. The extent of selection bias was determined by the magnitudes of genetic and nongenetic effects on the indexing (first) coronary heart disease event. Unless the genetic hazard ratio was unrealistically large (>1.6 per allele) and assuming the sum of all nongenetic hazard ratios was <10, bias was usually <10% (downward toward the null). Despite the low bias, the probability that a confidence interval included the true effect decreased (undercoverage) with increasing sample size because of increasing precision. Importantly, false-positive rates were not affected by selection bias. In most empirical settings, selection bias is expected to have a limited impact on genetic effect estimates of subsequent event risk. Nevertheless, because of undercoverage increasing with sample size, most confidence intervals will be over precise (not wide enough). When there is no effect modification by history of coronary heart disease, the false-positive rates of association tests will be close to nominal.
| S-EPMC5659743 | biostudies-literature
Project description:Background Tumour mutation burden (TMB), defined as the number of somatic mutations per megabase within the sequenced region in the tumour sample, has been used as a biomarker for predicting response to immune therapy. Several studies have been conducted to assess the utility of TMB for various cancer types; however, methods to measure TMB have not been adequately evaluated. In this study, we identified two sources of bias in current methods to calculate TMB. Methods We used simulated data to quantify the two sources of bias and their effect on TMB calculation, we down-sampled sequencing reads from exome sequencing datasets from TCGA to evaluate the consistency in TMB estimation across different sequencing depths. We analyzed data from ten cancer cohorts to investigate the relationship between inferred TMB and sequencing depth. Results We found that TMB, estimated by counting the number of somatic mutations above a threshold frequency (typically 0.05), is not robust to sequencing depth. Furthermore, we show that, because only mutations with an observed frequency greater than the threshold are considered, the observed mutant allele frequency provides a biased estimate of the true frequency. This can result in substantial over-estimation of the TMB, when the cancer sample includes a large number of somatic mutations at low frequencies, and exacerbates the lack of robustness of TMB to variation in sequencing depth and tumour purity. Conclusion Our results demonstrate that care needs to be taken in the estimation of TMB to ensure that results are unbiased and consistent across studies and we suggest that accurate and robust estimation of TMB could be achieved using statistical models that estimate the full mutant allele frequency spectrum. Supplementary Information The online version contains supplementary material available at (10.1186/s12885-022-09897-3).
| S-EPMC9347149 | biostudies-literature
Project description:Built environment research is dominated by cross-sectional designs, which are particularly vulnerable to residential self-selection bias resulting from health-related attitudes, neighborhood preferences, or other unmeasured characteristics related to both neighborhood choice and health-related outcomes.We used cohort data from the National Longitudinal Study of Adolescent Health (United States; Wave I, 1994-95; Wave III, 2001-02; n = 12,701) and a time-varying geographic information system. Longitudinal relationships between moderate to vigorous physical activity (MVPA) bouts and built and socioeconomic environment measures (landcover diversity, pay and public physical activity facilities per 10,000 population, street connectivity, median household income, and crime rate) from adolescence to young adulthood were estimated using random effects models (biased by unmeasured confounders) and fixed effects models (within-person estimator, which adjusts for unmeasured confounders that are stable over time).Random effects models yielded null associations except for negative crime-MVPA associations [coefficient (95% CI): -0.056 (-0.083, -0.029) in males, -0.061 (-0.090, -0.033) in females]. After controlling for measured and time invariant unmeasured characteristics using within-person estimators, MVPA was higher with greater physical activity pay facilities in males [coefficient (95% CI): 0.024 (0.006, 0.042)], and lower with higher crime rates in males [coefficient (95% CI): -0.107 (-0.140, -0.075)] and females [coefficient (95% CI): -0.046 (-0.083, -0.009)]. Other associations were null or in the counter-intuitive direction.Comparison of within-person estimates to estimates unadjusted for unmeasured characteristics suggest that residential self-selection can bias associations toward the null, as opposed to its typical characterization as a positive confounder. Differential environment-MVPA associations by residential relocation suggest that studies examining changes following residential relocation may be vulnerable to selection bias. The authors discuss complexities of adjusting for residential self-selection and residential relocation, particularly during the adolescent to young adult transition.
| S-EPMC2959083 | biostudies-other