# Reviewer response kit — single-cell and spatial analyses A working template for the five analysis requests reviewers actually make. Copy the relevant section, replace the bracketed parts with your numbers, and delete the rest. Free to use and to modify; no attribution required. Prepared by OMICSDESK — https://omics.sinogenomics.com --- ## 1. "The QC thresholds appear arbitrary" **What the reviewer wants:** evidence that your conclusions do not depend on where you drew the lines. **What to run:** re-cluster at two or three alternative threshold sets and compare cluster number and the marker genes of your key population. **Template response** > We thank the reviewer for this point. Quality-control thresholds were applied per sample using > [median ± 3 MAD / fixed thresholds of X genes and Y% mitochondrial reads]. To confirm that our conclusions are > not threshold-dependent, we repeated clustering under two alternative filtering schemes > ([scheme A] and [scheme B]). The number of clusters changed from [N] to [N'] and [N''], and the > [population of interest] was recovered in all cases with the same marker profile (new Supplementary Figure SX, > new Supplementary Table SY). Per-sample cell counts before and after filtering are now reported in > Supplementary Table SZ. --- ## 2. "How do you know these clusters are cell type X?" **What the reviewer wants:** more than one line of evidence. **What to run:** a reference-based classifier (SingleR, Azimuth, CellTypist) run blind to your manual labels, plus gene-set scores, compared cluster by cluster. **Template response** > Cell-type assignments were originally based on canonical markers. To provide independent support we additionally > annotated all cells with [classifier, version] against [reference, version], without reference to the manual > labels. The two assignments agree for [N] of [M] clusters at the lineage level (new Supplementary Figure SX); > within each cluster, [percentage]% of cells received the same automated label. Clusters [list] where the two > approaches disagreed are now labelled at the broader lineage level and discussed as such in the Results. --- ## 3. "Batch effects have not been addressed" **What the reviewer wants:** to know that batch was handled without erasing biology. **What to run:** integration, then a check that a population known to differ between conditions still separates. **Template response** > Samples were processed in [N] batches ([describe]). We integrated using [method, version] on [which variable]. > To verify that integration did not remove biological signal, we confirmed that [known condition-specific > population] remains separable after correction and that its marker profile is unchanged (new Supplementary > Figure SX). Differential expression was performed on uncorrected counts with batch included as a covariate, > rather than on integrated values. --- ## 4. "The differential expression is not statistically appropriate" **What the reviewer wants:** the statistical unit to be the subject, not the cell. **What to run:** pseudobulk aggregation per subject per cell type, then DESeq2 / edgeR / limma-voom. **Template response** > We agree that treating individual cells as independent replicates is not appropriate for a comparison between > groups of [animals / donors]. We have repeated the analysis using pseudobulk aggregation: raw counts were summed > across cells of each cell type within each [subject], giving n = [N] per group, and tested with [DESeq2 vX] with > [covariates] in the model. Samples contributing fewer than [threshold] cells to a given cell type were excluded > and are listed in Supplementary Table SX. The revised analysis identifies [N] differentially expressed genes > ([previous number] under the earlier per-cell test); [describe what changed]. All figures and text have been > updated accordingly. --- ## 5. "Please add trajectory / cell–cell communication / pathway analysis" **What the reviewer wants:** the additional module, framed honestly. **Template response** > We have added [module] using [tool, version] with [parameters]. [For trajectory: topology was first estimated > with PAGA and the pseudotime root was set to [population] based on [justification].] [For communication: results > were computed with [tool] against [database, version] and are reported as candidate interactions.] We note that > this analysis nominates hypotheses rather than demonstrating [progression / physical interaction], and the text > has been revised to reflect this. --- ## Before you resubmit — a short checklist - [ ] Per-sample QC table included (cells before/after, thresholds, % removed) - [ ] Doublet handling named, with the tool and the rate - [ ] Ambient RNA addressed or explicitly stated as checked and low - [ ] Statistical unit stated in the methods ("n refers to donors, not cells") - [ ] Tool versions and parameters in the methods - [ ] Code and processed object deposited or available on request - [ ] Every new supplementary figure referenced in the response letter --- If a request needs an analysis you cannot run in time, an independent re-run is usually faster than arguing about it: https://omics.sinogenomics.com — fixed price agreed before any work starts, and reviewer-driven revisions included for six months.