Biology subjects

Schwartzman, A.

Publications and source records attributed to Schwartzman, A..

13 recordsLinked to original sources

bioRxiv2026

Conditional and marginal SNP-heritability to leverage ancestral and environmental diversity

SNP-heritability is defined as the fraction of variance of a trait that is explained by the SNPs in a genome-wide association study. Several methodologies have been proposed to estimate this quantity. More recent methods aim to do so with ancestrally diverse datasets and yet obtain a single heritability for an entire dataset, which we refer to as marginal heritability. However, the different underlying subpopulations that compose a genetically diverse dataset might have different environmental and genetic exposures, and thus may have different heritabilities. In order to address this, we propose a conditional SNP-heritability approach that allows to estimate multiple SNP-heritabilities on a dataset corresponding to different ancestral compositions and environmental exposures. We take a careful statistical approach, including estimation of conditional genetic and environmental variances, and calculation of standard errors via a combination of the delta method with bootstrapping. We validate our method via extensive simulations. We then apply it to an ancestrally and socio-economically diverse dataset of 6603 subjects aged around 9 to 11 from the Adolescent Brain Cognitive Development study, and illustrate how the SNP-heritability of intelligence scores can change due to differing extrinsic variances in different socio-economic groups, which coincides with previous work in the literature. This conditional estimation approach can be a valuable tool for understanding differences in risks across subpopulations. Our work here improves on existing methodology and allows us to leverage the heterogeneity of the data to obtain new insights.

genetics↗

bioRxiv2026

Quantifying uncertainty in drift diffusion models of decision making under temporal dependence and parameter variability

Decision-making behavior changes over time, exhibiting temporal correlation and nonstationarity. Existing drift diffusion model (DDM) fitting methods either do not provide uncertainty quantification for parameter estimates, or rely on restrictive assumptions that decisions are independent and that parameters remain constant over time, potentially underestimating uncertainty. To address these limitations, we propose a computationally efficient method for estimating analytic uncertainties in DDM parameters that are robust to temporal dependence and unmodeled parameter variability, while explicitly modeling nonstationary variability through covariates. We apply this method to rat decision-making in a two-alternative forced-choice (2AFC) visual task, revealing dynamic decision-making states across multiple timescales. A Python implementation of the method is provided.

animal behavior and cognition↗

bioRxiv2026

Estimating the fraction of variance of crystallized intelligence explained by cortical surface area in early adolescence

The relationship between cortical morphology and intelligence during adolescence has been widely studied, with existing literature reporting varying degrees of association across different modeling approaches. This study provides a comprehensive comparison of model performance in investigating the association between crystallized intelligence and cortical surface area using data from 11,351 subjects in the Adolescent Brain Cognitive Development (ABCD) study. We evaluate ten widely used models ranging from linear regression to graph convolutional networks across three covariate adjustment formulations: full (no adjustment), partial (age and sex adjusted), and total surface area (TSA) partial (age, sex, and TSA adjusted). Using bootstrap resampling with 50 iterations, we estimate the fraction of variance explained (FVE) for each model. Our results suggest that more complex models do not lead to higher FVE, with LASSO having the highest FVE of 15.9% (full formulation), Ridge at 10.5% (partial formulation), and Principal Component Regression (PCR) with 102 PCs at 2.5% (TSA partial formulation). Our results also reveal that the relationship between cortical surface area and crystallized intelligence is predominantly driven by global factors age, sex, and TSA, rather than by localized cortical surface area.

neuroscience↗

bioRxiv2026

When can whole-genome SNP heritability be reliably estimated from summary statistics?

LD Score Regression (LDSC) is a prominent method, which estimates whole-genome SNP heritability from summary statistics via the slope of a linear regression of GWAS test statistics corresponding to a trait of interest against LD scores. It was claimed by the LDSC authors that the free intercept in the regression accounts for confounding bias such as population stratification. In this study, we argue that the intercept in LDSC must be fixed to 1 for accurate SNP heritability estimation. We show both theoretically and with simulations that the estimated intercept does not accurately capture population stratification effects, and that it adversely affects the accuracy of the heritability estimate introducing bias and increasing variance. Fixing the intercept to 1 eliminates bias and reduces variance when no population stratification is present. On the other hand, under population stratification, LDSC is biased with both the free and the fixed intercept. Additionally, we show that estimated standard errors in LDSC are underestimated, potentially leading to false-positives in downstream GWAS analyses.

genetics↗

bioRxiv2026

Partitioning Fraction of Variance Explained into Strong Localized Effects and Weak Diffuse Effects

High-dimensional genetic data present substantial challenges for estimating the fraction of variance explained (FVE) by genome-wide single-nucleotide polymorphisms (SNPs). In the context of genetics the VFE is called SNP heritability. Standard approaches for FVE estimation, such as GWAS heritability (GWASH) and linkage disequilibrium score (LDSC) regression, typically assume Gaussian distributions for SNP effect sizes. However, empirical evidence indicates that SNP effects are often heavy-tailed, with a small subset of variants exerting disproportionately large influence. Such settings violate the recently established bounded-kurtosis effect (BKE) condition, under which these FVE estimators are consistent. Consequently, widely used methods may yield severely biased estimates when strong effects are present. We introduce a decomposed FVE estimation framework that accommodates heavy-tailed and heterogeneous SNP effect distributions. The proposed approach partitions total heritability into contributions from strong and weak genetic effects, estimating the former using low-dimensional adjusted R2 and the latter using an extension of FVE estimation methodology that remains valid under BKE compliance. We further develop a test for detecting violations of the BKE condition and compare several high-dimensional screening procedures for identifying strong-effect SNPs when they are not known in advance. Simulation studies show that the proposed decomposition substantially improves estimation accuracy over existing approaches in the presence of heavy-tailed effects. Application to the Adolescent Brain Cognitive Development (ABCD) Study demonstrates the practical utility of the method, yielding more reliable heritability estimates for the PolyVoxel Score, a neuroimaging-based biomarker linked to iron accumulation. Our results highlight the importance of accommodating effect heterogeneity in large-scale genomic studies.

genetics↗

bioRxiv2025

Estimating fMRI Timescale Maps

Brain activity unfolds over hierarchical timescales that reflect how brain regions integrate and process information, linking functional and structural organization. While timescale studies are prevalent, existing estimation methods rely on the restrictive assumption of exponentially decaying temporal autocorrelation and only provide point estimates without standard errors, limiting statistical inference. In this paper, we formalize and evaluate two methods for mapping timescales in resting-state fMRI: a time-domain fit of an autoregressive (AR1) model and an autocorrelation-domain fit of an exponential decay model. Rather than assuming exponential autocorrelation decay, we define timescales by projecting the fMRI time series onto these approximating models, requiring only stationarity and mixing conditions while incorporating robust standard errors to account for model misspecification. We introduce theoretical properties of timescale estimators and show parameter recovery in realistic simulations, as well as applications to fMRI from the Human Connectome Project. Comparatively, the time-domain method produces more accurate estimates under model misspecification, remains computationally efficient for high-dimensional fMRI data, and yields maps aligned with known functional brain organization. In this work, we show valid statistical inference on fMRI timescale maps, and provide Python implementations of all methods.

neuroscience↗

bioRxiv2025

Simultaneous Confidence Regions for Image Excursion Sets: a Validation Study with Applications in fMRI

Functional Magnetic Resonance Imaging (fMRI) is commonly used to localize brain regions activated during a task. Methods have been developed for constructing confidence regions of image excursion sets, allowing inference on brain regions exceeding non-zero activation thresholds. However, these methods have been limited to a single predefined threshold and brain volume data, overlooking more sensitive cortical surface analyses. We present an approach that constructs simultaneous confidence regions (SCRs) which are valid for all possible activation thresholds and are applicable to both volume and surface data. This approach is based on a recent method that constructs SCRs from simultaneous confidence bands (SCBs), obtained by using the bootstrap on 1D and 2D images. To extend this method to fMRI studies, we evaluate the validity of the bootstrap with fMRI data through extensive 2D simulations. Six bootstrap variants, including the nonparametric bootstrap and multiplier bootstrap are compared. The Rademacher multiplier bootstrap-t performs the best, achieving a coverage rate close to the nominal level with sample sizes as low as 20. We further validate our approach using realistic noise simulations obtained by resampling resting-state 3D fMRI data, a technique that has become the gold standard in the field. Moreover, our implementation handles data of any dimension and is equipped with interactive visualization tools designed for fMRI analysis. We apply our approach to task fMRI volume data and surface data from the Human Connectome Project, showcasing the methods utility.

neuroscience↗

bioRxiv2023

Evaluation of resampling-based inference for topological features of neuroimages

Many recent studies have demonstrated the inflated type 1 error rate of the original Gaussian random field (GRF) methods for inference of neuroimages and identified resampling (permutation and bootstrapping) methods that have better performance. There has been no evaluation of resampling procedures when using robust (sandwich) statistical images with different topological features (TF) used for neuroimaging inference. Here, we consider estimation of distributions TFs of a statistical image and evaluate resampling procedures that can be used when exchangeability is violated. We compare the methods using realistic simulations and study sex differences in life-span age-related changes in gray matter volume in the Nathan Kline Institute Rockland sample. We find that our proposed wild bootstrap and the commonly used permutation procedure perform well in sample sizes above 50 under realistic simulations with heteroskedasticity. The Rademacher wild bootstrap has fewer assumptions than the permutation and performs similarly in samples of 100 or more, so is valid in a broader range of conditions. We also evaluate the GRF-based pTFCE method and show that it has inflated error rates in samples less than 200. Our R package, pbj, is available on Github and allows the user to reproducibly implement various resampling-based group level neuroimage analyses.

neuroscience↗

bioRxiv2021

Bayesian Multivariate Growth Mixture Modeling of Longitudinal Data: An Application to Alzheimer's Disease Study

Alzheimers disease (AD) studies often collect longitudinal biomarker measures of multiple cohorts at different stages of disease and follow these biomarkers with a relatively short period of time. The heterogeneity of the longitudinal patterns of biomarkers can be ubiquitous across both individual trajectories and cognitive domains. We propose a flexible Bayesian multivariate growth mixture model to identify distinct longitudinal patterns of data from the Alzheimers Disease Neuroimaging Initiative (ADNI) study. A Gibbs sampling is implemented for achieving the Bayesian inference. We perform a simulation study to demonstrate the adequate performance of our proposed approach and apply the model to identify three latent cognitive decline patterns among patients from the ADNI study.

neuroscience↗

bioRxiv2020

Spatial Confidence Sets for Standardized Effect Size Images

Current statistical inference methods for task-fMRI suffer from two fundamental limitations. First, the focus is solely on detection of non-zero signal or signal change, a problem that is exasperated for large scale studies (e.g. UK Biobank, N = 40, 000+) where the null hypothesis fallacy causes even trivial effects to be determined as significant. Second, for any sample size, widely used cluster inference methods only indicate regions where a null hypothesis can be rejected, without providing any notion of spatial uncertainty about the activation. In this work, we address these issues by developing spatial Confidence Sets (CSs) on clusters found in thresholded Cohens d effect size images. We produce an upper and lower CS to make confidence statements about brain regions where Cohens d effect sizes have exceeded and fallen short of a non-zero threshold, respectively. The CSs convey information about the magnitude and reliability of effect sizes that is usually given separately in a t-statistic and effect estimate map. We expand the theory developed in our previous work on CSs for %BOLD change effect maps (Bowring et al., 2019) using recent results from the bootstrapping literature. By assessing the empirical coverage with 2D and 3D Monte Carlo simulations resembling fMRI data, we find our method is accurate in sample sizes as low as N = 60. We compute Cohens d CSs for the Human Connectome Project working memory taskfMRI data, illustrating the brain regions with a reliable Cohens d response for a given threshold. By comparing the CSs with results obtained from a traditional statistical voxelwise inference, we highlight the improvement in activation localization that can be gained with the Confidence Sets.

neuroscience↗

bioRxiv2019

Spatial Confidence Sets for Raw Effect Size Images

The mass-univariate approach for functional magnetic resonance imagery (fMRI) analysis remains a widely used and fundamental statistical tool within neuroimaging. However, this method suffers from at least two fundamental limitations: First, with sample sizes growing to 4, 5 or even 6 digits, the entire approach is undermined by the null hypothesis fallacy, i.e. with sufficient sample size, there is high enough statistical power to reject the null hypothesis everywhere, making it difficult if not impossible to localize effects of interest. Second, with any sample size, when cluster-size inference is used a significant p-value only indicates that a cluster is larger than chance, and no notion of spatial uncertainty is provided. Therefore, no perception of confidence is available to express the size or location of a cluster that could be expected with repeated sampling from the population.\n\nIn this work, we address these issues by extending on a method proposed by Sommerfeld, Sain, and Schwartzman (2018) to develop spatial Confidence Sets (CSs) on clusters found in thresholded raw effect size maps. While hypothesis testing indicates where the null, i.e. a raw effect size of zero, can be rejected, the CSs give statements on the locations where raw effect sizes exceed, and fall short of, a non-zero threshold, providing both an upper and lower CS.\n\nWhile the method can be applied to any parameter in a mass-univariate General Linear Model, we motivate the method in the context of BOLD fMRI contrast maps for inference on percentage BOLD change raw effects. We propose several theoretical and practical implementation advancements to the original method in order to deliver an improved performance in small-sample settings. We validate the method with 3D Monte Carlo simulations that resemble fMRI data. Finally, we compute CSs for the Human Connectome Project working memory task contrast images, illustrating the brain regions that show a reliable %BOLD change for a given %BOLD threshold.

neuroscience↗

bioRxiv2018

Peak p-values and false discovery rate inference in neuroimaging

Peaks are a mainstay of neuroimage analysis for reporting localization results. The current peak detection procedure in SPM12 requires a pre-threshold for approximating p-values and a false discovery rate (FDR) nominal level for inference. However, the pre-threshold is an undesirable feature, while the FDR level is meaningless if the signal is assumed to be nonzero everywhere. This article provides: 1) a peak height distribution for smooth Gaussian error fields that does not require a screening pre-threshold; 2) a signal-plus-noise model where FDR of peaks can be controlled and properly interpreted. Matlab code for calculation of p-values using the exact peak height distribution is available as an SPM extension.

neuroscience↗

bioRxiv2017

A simple, consistent estimator of heritability for genome-wide association studies

Analysis of genome-wide association studies (GWAS) is characterized by a large number of univariate regressions where an outcome, a quantitative trait, is regressed on hundreds of thousands to millions of genomic markers, i.e. single-nucleotide polymorphism (SNP) counts, one marker at a time. Assuming a linear model linking the markers to the outcome, this article proposes an estimator of the heritability of the trait, defined here as the fraction of the variance of the trait explained by the genomic markers in the study. The estimator, called GWAS heritability (GWASH) estimator, is easy to compute, highly interpretable, and is consistent as the number of markers and the sample size increase. More importantly, it can be computed from summary statistics typically reported in GWAS, not requiring access to the original data. The estimator takes full account of the linkage disequilibrium (LD) or correlation between the SNPs in the study through moments of the LD matrix, estimable from auxiliary datasets. Unlike other proposed estimators in the literature, the precision of the estimate is obtainable analytically, allowing for power and sample size calculations for heritability estimates.

bioinformatics↗