#REML
LightNews LightNews
paperposterbot.bsky.social
arXiv📈🤖
Improving the selection-function-based sensitivity analysis for publication bias with restricted maximum likelihood
By Hu, Zhou, Zhou et al
In random-effects meta-analysis, the between-study variance quantifies the heterogeneity among studies. Without selective publication, the restricted maximum likelihood (REML) method is recommended for this variance, since the maximum likelihood (ML) method underestimates it when the number of studies is small. However, under selective publication, referred to as publication bias (PB), the published studies no longer follow the normal-normal model, and the REML cannot be directly applied. Thus, the existing sensitivity analysis methods for PB based on selection functions estimate the between-study variance by the ML conditional on published studies and inherit its underestimation. This underestimation also affects the estimate of the common effect and its confidence interval. To improve the estimation of the common effect and the heterogeneity, we propose a REML-type estimator of the between-study variance for the selection-function-based sensitivity analysis. It centers the profile score by its approximate mean and reduces exactly to the REML without PB. The proposed sensitivity analysis method is a more accurate alternative to the ML-based sensitivity analysis in meta-analyses with a small number of studies, and it is the first REML-type estimator in the presence of PB. Simulation studies substantiate that the proposed method reduces the median bias of the between-study variance and improves the estimation error and coverage probability for the common effect compared with the conditional ML. We illustrate the proposed method with two meta-analyses on high maintenance-dose clopidogrel and nicotine gum for smoking cessation. The proposed method would give useful guidance on the between-study heterogeneity in sensitivity analysis for PB.
November 4, 2026 at 4:36 AM
arxivecongnbot.bsky.social
arXiv📈🤖
Restricted maximum likelihood estimation in generalized linear mixed models
By
arXiv:2402.12719v1 Announce Type: new Abstract: Restricted maximum likelihood (REML) estimation is a widely accepted and frequently used method for fitting linear mixed models, with its principal advantage being that it produces less biased estimates of the variance components. However, the concept of REML does not immediately generalize to the setting of non-normally distributed responses, and it is not always clear the extent to which, either asymptotically or in finite samples, such generalizations reduce the bias of variance component estimates compared to standard unrestricted maximum likelihood estimation. In this article, we review various attempts that have been made over the past four decades to extend REML estimation in generalized linear mixed models. We establish four major classes of approaches, namely approximate linearization, integrated likelihood, modified profile likelihoods, and direct bias correction of the score function, and show that while these four classes may have differing motivations and derivations, they often arrive at a similar if not the same REML estimate. We compare the finite sample performance of these four classes through a numerical study involving binary and count data, with results demonstrating that they perform similarly well in reducing the finite sample bias of variance components.
November 2, 2026 at 11:19 PM
icyhawaiian.bsky.social
The MLMA step of a GWAS can be slow, so chunking variants into smaller subsets can help parallelise. But MLMA only runs after REML, so each chunk had to (redundantly) run its own REML.

The simple fix is just to serialise the REML state once and load that into each parallel MLMA chunk.

Job done?
October 29, 2026 at 10:12 AM
icyhawaiian.bsky.social
We can also do some out-of-core GRM operations while building this reduced basis, reducing the optimised REML step on 24 cores from ~268 GB and 14 hours to ~54 GB and 1 hour — with under 5% genomic inflation.
GCTAg approximate REML benchmarks
October 29, 2026 at 10:12 AM
icyhawaiian.bsky.social
So by addressing several memory bottlenecks (and some other optimisations), we can actually run end-to-end GWAS on smaller machines.

For N=75k, we need around 245 GB of memory for GCTA (peaks during REML), but I can run the same dataset on my 24 GB laptop now with minor genomic inflation.
October 29, 2026 at 10:12 AM
cranupdates.bsky.social
New on CRAN: cepip (0.1.0). View at https://CRAN.R-project.org/package=cepip
cepip: Cell Explanatory Power with Inflection Point (CEP-IP) with CMRF
Implements the cell explanatory power with inflection point (CEP-IP) framework for single-cell co-expression analysis. A generalized additive model (GAM) fitted by restricted maximum likelihood (REML) with a penalized residual sum of squares (PRSS) search over the basis dimension relates a gene of interest to the mean of its partner genes. Cells are ranked by their explanatory power, and the inflection point is the first negative-to-positive zero crossing of a residual GAM. Optional extensions are the Spearman-Kendall cell-level score (SKCS) and the copula monotone recovery filter (CMRF), which recovers cells near the module through a tolerance band and a repeated stratified k-fold cross-validation gate. Downstream helpers give per-donor module fractions, paired donor-level differential expression and donor-level association models. CEP-IP is described in Wong (2026) &lt;<a href="https://doi.org/10.1016%2Fj.cmpb.2026.109372" target="_top">doi:10.1016/j.cmpb.2026.109372</a>&gt;.
CRAN.R-project.org
October 29, 2026 at 9:31 PM
cranberriesfeed.bsky.social
New CRAN package cepip with initial version 0.1.0
#rstats
https://cran.r-project.org/package=cepip
CRAN: Package cepip
Implements the cell explanatory power with inflection point (CEP-IP) framework for single-cell co-expression analysis. A generalized additive model (GAM) fitted by restricted maximum likelihood (REML) with a penalized residual sum of squares (PRSS) search over the basis dimension relates a gene of interest to the mean of its partner genes. Cells are ranked by their explanatory power, and the inflection point is the first negative-to-positive zero crossing of a residual GAM. Optional extensions are the Spearman-Kendall cell-level score (SKCS) and the copula monotone recovery filter (CMRF), which recovers cells near the module through a tolerance band and a repeated stratified k-fold cross-validation gate. Downstream helpers give per-donor module fractions, paired donor-level differential expression and donor-level association models. CEP-IP is described in Wong (2026) &lt;<a href="https://doi.org/10.1016%2Fj.cmpb.2026.109372" target="_top">doi:10.1016/j.cmpb.2026.109372</a>&gt;.
cran.r-project.org
October 29, 2026 at 8:02 PM
spletters.bsky.social
❓Can restricted maximum likelihood estimation for spatial models be made computationally scalable?

👉Debjoy Thakur (Ahmedabad University and Washington University in St. Louis) develops a variational approximation to REML for spatial data

Read the letter here⤵️
www.sciencedirect.com/science/arti...
www.sciencedirect.com
October 20, 2026 at 5:12 PM
cranberriesfeed.bsky.social
New CRAN package hsquared with initial version 0.9.0
#rstats
https://cran.r-project.org/package=hsquared
CRAN: Package hsquared
Provides an interface for quantitative-genetic models in 'R', backed by the 'HSquared.jl' engine for 'Julia'. Fits Gaussian pedigree animal models by restricted maximum likelihood, using average-information restricted maximum likelihood (AI-REML; Gilmour et al. (1995) &lt;<a href="https://doi.org/10.2307%2F2533274" target="_top">doi:10.2307/2533274</a>&gt;) for the univariate model. Returns heritability, variance components, breeding values, and covariance and correlation matrices for multivariate models. Formula and pedigree validation can run without 'Julia'. Model fitting requires a local 'Julia' and 'HSquared.jl'; use hs_control(engine = "validate") to preview the model contract without fitting. What you may report is listed on the package article 'Can I fit and report this?', not in validation_status(), which is a developer evidence table. Model-specific restrictions are documented in that article. Uncertainty intervals are not coverage-calibrated.
cran.r-project.org
October 21, 2026 at 2:02 PM
cranupdates.bsky.social
New on CRAN: hsquared (0.9.0). View at https://CRAN.R-project.org/package=hsquared
hsquared: 'R' Interface for 'Julia'-Backed Quantitative-Genetic Models
Provides an interface for quantitative-genetic models in 'R', backed by the 'HSquared.jl' engine for 'Julia'. Fits Gaussian pedigree animal models by restricted maximum likelihood, using average-information restricted maximum likelihood (AI-REML; Gilmour et al. (1995) &lt;<a href="https://doi.org/10.2307%2F2533274" target="_top">doi:10.2307/2533274</a>&gt;) for the univariate model. Returns heritability, variance components, breeding values, and covariance and correlation matrices for multivariate models. Formula and pedigree validation can run without 'Julia'. Model fitting requires a local 'Julia' and 'HSquared.jl'; use hs_control(engine = "validate") to preview the model contract without fitting. What you may report is listed on the package article 'Can I fit and report this?', not in validation_status(), which is a developer evidence table. Model-specific restrictions are documented in that article. Uncertainty intervals are not coverage-calibrated.
CRAN.R-project.org
October 21, 2026 at 2:17 PM
maxmax.bsky.social
Obligatory. bsky.app/profile/reml...
remlyricshourly.bsky.social Hourly R.E.M. Lyrics @remlyricshourly.bsky.social · Oct 8
You're sharpening stones, walking on coals
To improve your business acumen.
Sharpening stones, walking on coals,
To improve your business acumen.

Song: Exhuming McCarthy
Album: Document
Exhuming McCarthy at Youtube
Exhuming McCarthy at Youtube
www.youtube.com
October 2, 2026 at 11:11 AM
maxmax.bsky.social
bsky.app/profile/reml...
remlyricshourly.bsky.social Hourly R.E.M. Lyrics @remlyricshourly.bsky.social · Jul 21
They called the clip a two-headed cow
Your hate clipped and distant, your luck, pilgrimage
Rest assured this will not last, take a turn for the worst
Your hate clipped and distant, your luck a two-headed cow
The pilgrimage has gained momentum
...
Song: Pilgrimage
Album: Murmur
Pilgrimage at Youtube
Pilgrimage at Youtube
www.youtube.com
September 30, 2026 at 10:56 PM
damillora.com
The 5th LIVE visual for Reml has been revealed as well.
September 30, 2026 at 1:00 PM
paper.migraine.kr
두 명의 검토자가 독립적으로 데이터를 추출하고 ROBINS-I를 사용하여 비뚤림 위험을 평가하였다. 연속형 결과는 표준화 평균 차이(SMD; Hedges g; 랜덤 효과 REML)로 통합되었으며, VM 코호트에서 연구별로 정의된 인지 기능 저하의 유병률은 Freeman-Tukey 변환을 사용하여 종합하였다. 필요한 경우, 중앙값과 분산 측정값은 평균과 표준 편차로 변환되었다. (🧵 2/6)
September 18, 2026 at 6:42 PM
research.migraine.kr
Two reviewers independently extracted data and assessed the risk of bias using ROBINS-I. Continuous outcomes were pooled as standardized mean differences (SMD; Hedges g; random-effects REML), while the prevalence of study-defined cognitive impairment in VM cohorts was synthesized (🧵 3/9)
September 18, 2026 at 5:11 PM
paperposterbot.bsky.social
arXiv📈🤖
Asymptotic Theory for Restricted Maximum Likelihood Estimators in Generalized Linear Mixed Models
By Gao, Mo, Lyu
Developing an asymptotic theory for restricted maximum likelihood (REML) estimators in generalized linear mixed models (GLMMs) has remained an open problem for decades. In this paper, we establish the asymptotic distributions of the REML and maximum likelihood (ML) estimators for GLMMs when both the number of clusters and the cluster sizes tend to infinity. Under very mild conditions, requiring only finite-moment assumptions on the random effects rather than normality and imposing no restriction on the relative rates at which the number of clusters and the cluster sizes diverge, both estimators are proved to be asymptotically normal with an explicit block-diagonal covariance structure, and the bias-correction property of REML is established. Simulation studies support the asymptotic results, and a genetic data analysis illustrates the methodology.
September 16, 2026 at 12:52 AM
paperposterbot.bsky.social
arXiv📈🤖
Asymptotic Theory for Restricted Maximum Likelihood Estimators in Generalized Linear Mixed Models
By Gao, Mo, Lyu
Developing an asymptotic theory for restricted maximum likelihood (REML) estimators in generalized linear mixed models (GLMMs) has remained an open problem for decades. In this paper, we establish the asymptotic distributions of the REML and maximum likelihood (ML) estimators for GLMMs when both the number of clusters and the cluster sizes tend to infinity. Under very mild conditions, requiring only finite-moment assumptions on the random effects rather than normality and imposing no restriction on the relative rates at which the number of clusters and the cluster sizes diverge, both estimators are proved to be asymptotically normal with an explicit block-diagonal covariance structure, and the bias-correction property of REML is established. Simulation studies support the asymptotic results, and a genetic data analysis illustrates the methodology.
September 15, 2026 at 8:19 PM
paperposterbot.bsky.social
arXiv📈🤖
Asymptotic Theory for Restricted Maximum Likelihood Estimators in Generalized Linear Mixed Models
By Gao, Mo, Lyu
Developing an asymptotic theory for restricted maximum likelihood (REML) estimators in generalized linear mixed models (GLMMs) has remained an open problem for decades. In this paper, we establish the asymptotic distributions of the REML and maximum likelihood (ML) estimators for GLMMs when both the number of clusters and the cluster sizes tend to infinity. Under very mild conditions, requiring only finite-moment assumptions on the random effects rather than normality and imposing no restriction on the relative rates at which the number of clusters and the cluster sizes diverge, both estimators are proved to be asymptotically normal with an explicit block-diagonal covariance structure, and the bias-correction property of REML is established. Simulation studies support the asymptotic results, and a genetic data analysis illustrates the methodology.
September 15, 2026 at 6:46 AM
arxivecongnbot.bsky.social
arXiv📈🤖
confintROB Package: Confindence Intervals in robust linear mixed models
By
arXiv:2404.08426v1 Announce Type: new Abstract: Statistical inference is a major scientific endeavor for many researchers. In terms of inferential methods implemented to mixed-effects models, significant progress has been made in the R software. However, these advances primarily concern classical estimators (ML, REML) and mainly focus on fixed effects. In the confintROB package, we have implemented various bootstrap methods for computing confidence intervals (CIs) not only for fixed effects but also for variance components. These methods can be implemented with the widely used lmer function from the lme4 package, as well as with the rlmer function from the robustlmm package and the varComprob function from the robustvarComp package. These functions implement robust estimation methods suitable for data with outliers. The confintROB package implements the Wald method for fixed effects, whereas for both fixed effects and variance components, two bootstrap methods are implemented: the parametric bootstrap and the wild bootstrap. Moreover, the confintROB package can obtain both the percentile and the bias-corrected accelerated versions of CIs.
September 10, 2026 at 9:30 PM
andrejpaleo.bsky.social
We all know big-headed giant theropods. But what about big-clawed and strong-arm predatory dinosaur evolution?
"..Megaraptoridae...the only large-bodied carnivorous theropod lineage retaining a specialized forelimb-based predatory strategy.."
link.springer.com/article/10.1...
🧪⚒️ #EevoBio #Paleobio
(A) Ancestral state reconstruction of log-transformed claw centroid size under a Brownian motion model estimated using REML. (B) Phenogram of log-transformed claw centroid size through time, illustrating the evolutionary trajectories of Megaraptora (blue), phylogenetic outgroups (light green), Allosauria (black), Megalosauria (green), and Coelurosauria (red). (C) Fitted values from the phylogenetic generalized least squares (PGLS) model relating Procrustes-aligned shape data to systematic category and size Phylogenetic integration patterns (effec t siz e in the left and r-PLS on the right of the black line) and evolutionary rates of claw, ulna-radio, and humerus among non-avian theropod clades with well-developed forelimbs. ( Joaquinraptor skeletal reconstruction modified from Ibiricu et al., 2025 ). Evolutionary rates of hindlimb elements were estimated using the set of Procrustes Coordinates ( compare.evol.rates function, geomorph v. 4.0.9 package) under a Brownian evolution model with 10,000 interactions and a permutational procedure.
September 7, 2026 at 12:24 PM
reinoirasuto.bsky.social
no its adachi rei reml vflower ui kafu kaai yuki hatsune miku and kasane teto (left to right)
August 28, 2026 at 6:18 PM
iforgot.bsky.social
youtube.com/watch?v=Reml...
【ゲスト：三宅健太】ソー＆キャプテン・アメリカ役の声優達と語らうMCU。御自慢のグッズもお見せしちゃいます！【ソー：ラブ＆サンダー】
YouTube video by マフィア梶田と中村悠一の「わしゃがなTV」
youtube.com
August 25, 2026 at 11:33 PM
epigenci.bsky.social
Some cute ideas for running REML faster are implemented.
August 20, 2026 at 1:01 AM
nichromekneesocks.bsky.social
Still the top result for "how to draw vtuber" on YouTube almost half a decade later. All according to keikaku, Ina.

#vtubertech #krita #digitalart #tutorial #hololive

www.youtube.com/watch?v=rEMl...
【DRAWING】 Let's Learn Krita Together!!
YouTube video by Ninomae Ina'nis Ch. hololive-EN
www.youtube.com
August 9, 2026 at 10:03 PM
kyara3usi9.bsky.social
I didn't realize this until now but the artist for REML is Mika Pikazo, who is the main artist for Fire Emblem Engage!
REML Fire Emblem Engage
February 16, 2026 at 2:24 PM