scLASER: a robust framework for simulating and detecting time-dependent single-cell dynamics in longitudinal studies

Name: Verified Rigor Record — scLASER: a robust framework for simulating and detecting time-dependent single-cell dynamics in longitudinal studies
Item: scLASER: a robust framework for simulating and detecting time-dependent single-cell dynamics in longitudinal studies
Rating: 95
Author: Αlpha¹

Vanderlinden, L. A.; Vargas, J.; Inamo, J.; Young, J.; Wang, C.; Zhang, F.

Paper source

Read PDF

Αlpha¹ Rigor Review

Rigor Review Verified

Critical

High-priority

Action items to resolve · Critical / High

Confidence: moderateEngine: 7.6.0Source: full text

Not verified in this report

No reported statistical tests were found to recompute.

This Alpha1 Rigor Review uses AI Rigor Reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.

The paper introduces scLASER, a computational framework for longitudinal single-cell data analysis, validated through simulations and two real-world datasets. The scientific premise, data/code availability, and reporting transparency are strong, but the study design and statistical reporting have notable gaps, including missing outlier handling, an internal cell-count discrepancy, lack of demographic summaries, and unverified statistical assumptions.

All three reviewers consistently rated the paper as a computational study. The synthesis weighed all three reviewers' evidence; the internal contradiction flagged by the integrity verification component was incorporated into study design. The statistics verification component had 0 coverage, meaning no statistical tests could be independently recomputed.

7 major claims checked against the paper's own evidence: all adequately supported.

supportedResults, 'scLASER outperforms cluster-based frequency approaches...'Reviewers 1, 2
scLASER shows consistently higher sensitivity than traditional cluster-based approaches.
Simulation results directly compare sensitivity (97.5% vs. 94.4% overall, gains in rare cell types).
Evidence: Results section reports sensitivity comparisons from 200 simulated datasets.
“Across 200 randomly generated datasets... scLASER achieved higher sensitivity than the cluster-based test (97.5% vs 94.4%).”
supportedResults, 'scLASER identifies treatment-responsive NOTCH3+ stromal populations in IBD'Reviewers 1, 2
Applications to IBD reveal treatment-responsive NOTCH3+ stromal trajectories with high cell type discrimination (AUC > 0.92).
The paper reports AUC values of 0.953, 0.950, and 0.926 for predicting remission based on NOTCH3+ populations.
Evidence: Results section provides specific AUC values.
“Predictive modeling using abundances of three NOTCH3+ populations achieved strong discrimination of remission status (AUC=0.953, 0.950, and 0.926, respectively).”
supportedResults, 'Three distinct T cell activation axes emerge during COVID-19 progression using scLASER'Reviewer 1
Analysis of COVID-19 data identifies three distinct axes of T cell activity over disease progression.
Three NAM PCs are described with associated gene programs and LRT p-values, supported by gene set enrichment.
Evidence: Results section ('Three distinct T cell activation axes...') describes three axes and their p-values.
“With scLASER, we identified three activation axes associated with temporal abundance shifts between progressors and non-progressors.”
supportedAbstract and DiscussionReviewers 1, 2, 3
scLASER enables robust longitudinal single-cell analysis and optimization of study design.
The framework is demonstrated through extensive simulations and real-data applications, including power estimation capability.
Evidence: The entire paper provides evidence through simulations and applications.
“scLASER enables robust longitudinal single-cell analysis and optimization of study design.”
supportedAbstract/ResultsReviewers 2, 3
Analysis of COVID-19 data identifies three distinct axes of T cell activity (cytotoxic effector, NK immunoreceptor signaling, and interferon-stimulated gene programs) over disease progression.
The paper describes three axes with associated LRT p-values and gene set enrichment, supporting the claim.
Evidence: Results: 'we identified three activation axes associated with temporal abundance shifts... Axis 1... Axis 2... Axis 3...' with LRT p-values 7e-04, 1e-04, and 7e-04.
“analysis of COVID-19 data (188,181 cells, 84 patients) identifies three distinct axes of T cell activity (cytotoxic effector, NK immunoreceptor signaling, and interferon-stimulated gene programs) over disease progression.”
supportedAbstract and ResultsReviewer 3
scLASER shows consistently higher sensitivity than traditional cluster-based approaches, with particularly pronounced gains in rare cell types and non-linear temporal patterns.
Supported by simulation benchmark results comparing sensitivity across 200 datasets.
Evidence: Simulation results report scLASER sensitivity of 97.5% vs 94.4% for cluster-based, and 96.7% vs 88.8% for rare types; for non-linear patterns, 87.8% vs 30.9%.
“scLASER shows consistently higher sensitivity than traditional cluster--based approaches, with particularly pronounced gains in rare cell types and non-linear temporal patterns.”
supportedResults and AbstractReviewer 3
Applications to inflammatory bowel disease reveal treatment-responsive NOTCH3+ stromal trajectories with high cell type discrimination (AUC > 0.92).
Supported by results showing LRT p=3e-4 for time-by-remission association and AUCs of 0.953, 0.950, 0.926 for three NOTCH3+ populations.
Evidence: Results: 'LRT p=3e-04' and 'Predictive modeling using abundances of three NOTCH3+ populations achieved strong discrimination of remission status (AUC=0.953, 0.950, and 0.926, respectively).'
“Applications to inflammatory bowel disease (95,813 cells, 38 patients) reveal treatment-responsive NOTCH3+ stromal trajectories with high cell type discrimination (AUC > 0.92)”

1 integrity concern flagged (0 high).

Rigor

Eight Dimensions6 pass · 1 warn

7 major claims7 supported

4 copyedit issues flagged: mostly other, typo.

Checked 15 references: 10 verified — 5 not checked.

Reproducibility

3 data/code links checked; 3 live.

dataZenodoLIVEHTTP 200
https://doi.org/10.5281/zenodo.13768607
Resolves to Zenodo (data repository).
dataZenodoLIVEHTTP 200
https://doi.org/10.5281/zenodo.5153528
Resolves to Zenodo (data repository).
codeGitHubLIVEHTTP 200
https://github.com/fanzhanglab/scLASER
Resolves to GitHub (code repository).

Alpha¹ Rigor Action ItemsFinal pass before journal submission

As a post-publication audit, the paper is moderately robust but an informed reader should weigh the unresolved cell-count discrepancy (95,813 vs 103,503 cells in the IBD dataset), the missing demographic characteristics of the study populations, and the lack of verified statistical assumptions (e.g., normality, homoscedasticity for LMMs) and confidence intervals for reported AUCs. These gaps do not invalidate the main findings but weaken the paper's transparency and could warrant a correction or independent re-analysis of the affected numbers.

1.
HIGHrigor
Resolve the internal contradiction in the number of IBD stromal cells: the Abstract/Results state 95,813 cells while the Methods state 103,503 cells; verify the correct number and correct all instances.
An unexplained numerical discrepancy undermines data integrity and would be flagged by an editor or reviewer.
2.
HIGHreporting
Add a summary table of participant demographics (age distribution, sex proportions, baseline disease severity, race/ethnicity if available) for both the IBD and COVID-19 cohorts in the Results or Methods section.
Transparency about study populations is essential for assessing generalizability; the current reporting lists only disease subtype and participant count.
3.
HIGHstatistics
Report confidence intervals for all effect sizes, including AUC values and any model coefficients reported in the Results.
Effect sizes without confidence intervals provide an incomplete picture of precision and are a standard reporting requirement.
4.
HIGHstatistics
Verify and report statistical assumptions (e.g., normality of residuals, homoscedasticity) for the linear mixed models used in the analysis, or justify why they are not required.
Unverified assumptions can undermine the validity of parametric test results (LRT, p-values).
5.
HIGHreporting
Describe how outliers were handled in the real-data analyses (e.g., exclusion criteria, robust methods) in the Materials and Methods section.
Outlier handling is a standard design element; missing this information reduces reproducibility.
6.
HIGHstatistics
Specify version numbers for all software used (R, scLASER, Harmony, ComBat) in the Methods section.
Version information is critical for computational reproducibility.
7.
MEDIUMdata code
Deposit the scLASER code in a long-term archive (e.g., Zenodo) and obtain a persistent DOI to supplement the GitHub URL.
A GitHub URL may become stale; a permanent identifier ensures long-term accessibility and is preferred by many journals.
8.
MEDIUMreporting
Add a brief statement on how missing data were handled in the real datasets (e.g., dropouts, intermittent missing values) to the Methods.
Missing data handling is a common source of bias and should be transparent.
9.
MEDIUMstatistics
Include a power analysis or sample size justification for the real-data applications (IBD, COVID-19 cohorts) as a supplement or note in the Discussion.
Even for method development, understanding the power to detect effects in real data contextualizes the results.
10.
LOWcopyedit
Replace the double dash '--' with a single hyphen in 'cluster--based' in the Abstract and apply this fix throughout the manuscript.
Typographical errors detract from manuscript polish.
11.
LOWcopyedit
Remove the duplicate URL in the Code Availability section.
Clean formatting is expected in a submission-ready manuscript.

Full text (36,127 chars)Analyzed: Jul 6, 2026

Αlpha¹ assesses methodological rigor, not the importance of the findings. See how we calibrate →