Differential abundance testing on single-cell data using k-nearest neighbor graphs
Open accessPublished version of a preprint
Abstract
Current computational workflows for comparative analyses of single-cell datasets typically use discrete clusters as input when testing for differential abundance among experimental conditions. However, clusters do not always provide the appropriate resolution and cannot capture continuous trajectories. Here we present Milo, a scalable statistical framework that performs differential abundance testing by assigning cells to partially overlapping neighborhoods on a k-nearest neighbor graph. Using simulations and single-cell RNA sequencing (scRNA-seq) data, we show that Milo can identify perturbations that are obscured by discretizing cells into clusters, that it maintains false discovery rate control across batch effects and that it outperforms alternative differential abundance testing strategies. Milo identifies the decline of a fate-biased epithelial precursor in the aging mouse thymus and identifies perturbations to multiple lineages in human cirrhotic liver. As Milo is based on a cell–cell similarity structure, it might also be applicable to single-cell data other than scRNA-seq. Milo is provided as an open-source R software package at https://github.com/MarioniLab/miloR. Milo identifies differentially abundant populations of cells in scRNA-seq data without clustering.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 22 cited sources checked · 0 supported · 0 flagged
- Findingown data
Milo performs better than other differential abundance testing strategies.
“and that it outperforms alternative differential abundance testing strategies”
Quote in the paperSubstantial claim - Findingown data
Milo keeps false discovery rate under control when batch effects are present.
“that it maintains FDR control across batch effects”
Quote in the paperSubstantial claim - Findingown data
Milo detects the loss of a fate-biased epithelial precursor in the ageing mouse thymus.
“Milo identifies the decline of a fate-biased epithelial precursor in the ageing mouse thymus”
Quote in the paperSubstantial claim - Findingown data
Milo detects perturbations across several lineages in human cirrhotic liver.
“and identifies perturbations to multiple lineages in human cirrhotic liver”
Quote in the paperSubstantial claim - Findingown data
Milo can detect perturbations that are hidden when cells are discretised into clusters.
“Using simulations and scRNA-seq data, we show that Milo can identify perturbations that are obscured by discretising cells into clusters”
Quote in the paperSubstantial claim
- Findingown data
MELD systematically underestimates the fold-change of true differentially abundant regions.
“We found that MELD consistently underestimates the fold-change of true DA regions (Supp Fig 7A-B).”
Quote in the paperSubstantial claim - Findingown data
Milo recovered all previously identified differentially abundant states, including changes in the sTEC population, which contained only 24 cells.
“We compared our results to those generated in the original publication, which demonstrated that we were able to identify all previously identified DA states (Fig 4D), including changes in the abundance of the ‘sTEC’ population, which consisted of just 24 cells.”
Quote in the paperSubstantial claim - Findingown data
The variability in Milo's statistical power is accounted for by the proportion of true positive cells that lie near the threshold used to define ground truth differential abundance.
“Variability in Milo power is explained by the fraction of true positive cells close to the DA threshold for definition of ground truth.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
- Findingown data
Differential gene expression restricted to disease-specific cholangiocytes detected over-expression of genes linked to fibrosis, wound healing and angiogenesis.
“Performing differential gene expression analysis restricted to this subset, we detected over-expression of genes associated with fibrosis, wound healing and angiogenesis (Supp Fig 15C-D)”
Quote in the paperSubstantial claim- Rests on an assumption: The gene signatures associated with the ductular reaction in human cirrhosis are largely unexplored.
- Findingown data
DA-seq had consistently lower sensitivity, even when simulated fold changes were large.
“In contrast, DA-seq showed a consistently lower sensitivity, even for large simulated fold changes (Fig 2C, Supp Fig4).”
Quote in the paperSubstantial claim - Findingown data
Milo resolved disease-specific subpopulations in hepatic endothelial cells at higher resolution than clustering-based analysis could.
“we first focused on the hepatic endothelial cells, where we resolved disease specific subpopulations at higher resolution than was possible by clustering-based analysis (Fig 5D).”
Quote in the paperSubstantial claim - Findingown data
Milo recovered differentially abundant neighbourhoods in every cluster that the original study identified as differentially abundant between cirrhotic and uninjured tissue.
“Milo recovered DA neighbourhoods in all clusters identified as differentially abundant between cirrhotic or uninjured tissue in the original study (Fig 5D).”
Quote in the paperSubstantial claim - Findingown data
Milo found multiple groups of neighbourhoods within the same pre-defined subclusters that changed in opposite directions between control and cirrhotic liver samples.
“Moreover, Milo identified multiple groups of neighbourhoods within the same pre-defined subclusters that showed opposing directions of differential abundance between the control and cirrhotic liver experimental samples (Fig 5D).”
Quote in the paperSubstantial claim - Findingown data
In the discrete cluster simulation, several methods had inflated false discovery rates because of compositional biases, most clearly when the differentially abundant cluster made up at least half of the dataset cells.
“In the discrete cluster simulation, we found that multiple methods had an inflated FDR due to compositional biases; this was most evident where the differentially abundant cluster comprised ≥50% of the dataset cells (Supp Fig 4A).”
Quote in the paperSubstantial claim - Findingown data
In cirrhosis-enriched neighbourhoods, Milo recovered over-expression of known markers of scar-associated endothelium, including ACKR1 and PLVAP.
“In the cirrhosis-enriched neighbourhoods, we recovered over-expression of known markers of scar-associated endothelium, including ACKR1 and PLVAP (Fig 5F) [2].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [2] Ramachandran, P. et al. Resolving the fibrotic niche of human liver cirrhosis at single-cell level. Nature 575, 512–518 (2019). source
Not checked in the free run.
- Findingown data
On simulated data built from the real KNN graph, Milo outperformed every other method tested.
“While in specific simulated datasets other methods achieve comparable performance, Milo out-performed all methods on simulated data generated using the real KNN graph.”
Quote in the paperSubstantial claim - Findingown data
Methods such as MELD that cannot include the experimental design in likelihood estimation perform much worse when batch effects are uncorrected.
“The inability of methods such as MELD to incorporate the experimental design in the likelihood estimation severely hinders performance in the presence of uncorrected batch effects (Supp Fig 7C-F).”
Quote in the paperSubstantial claim- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
- Findingown data
Milo detected strong differential abundance between healthy and cirrhotic cells in the cholangiocyte compartment, a lineage that the original study did not explore.
“Milo also identified strong DA between healthy and cirrhotic cells in lineages that were unexplored in the original study, such as the cholangiocyte compartment (Fig 5D).”
Quote in the paperSubstantial claim- Rests on an assumption: The gene signatures associated with the ductular reaction in human cirrhosis are largely unexplored.
- Findingown data
Modelling the batch effect explicitly as a regression term in Milo's GLM preserves the sensitivity of differential abundance detection.
“By contrast, explicit modelling of the batch effect as a regression term in the Milo GLM maintains the sensitivity of the differential abundance (Supp Fig 11).”
Quote in the paperSubstantial claim- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
- Findingown data
Total memory use across the Milo workflow scaled mainly with input dataset size, suggesting that the complexity and composition of the single cells largely determine memory requirements.
“Moreover, the total memory usage across all steps of the Milo workflow scaled primarily with the size of the input dataset (Fig 3B), indicating that the complexity and composition of the single-cells largely determines the memory requirements (Supp Fig 12).”
Quote in the paperSubstantial claim - Findingown data
Differential expression analysis between cells in differentially abundant neighbourhoods with positive and negative log fold changes identified 788 differentially expressed genes at a 5% false discovery rate.
“we performed differential expression analysis between cells in DA neighbourhoods with positive and negative log fold changes, identifying 788 differentially expressed genes (FDR 5%; Methods) (Fig 5F).”
Quote in the paperSubstantial claim - Findingown data
Applying in silico batch correction preserves the sensitivity of all differential abundance methods, comparable to no batch effect, across different batch effect magnitudes.
“We confirm that using an in silico batch correction maintains the sensitivity of all DA methods, comparable to the case of no batch effect, across different batch effect magnitudes (Supp Fig 9; batch effects corrected using fastMNN [19]”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [19] Haghverdi, L., Lun, A. T. L., Morgan, M. D. & Marioni, J. C. Batch effects in single-cell RNA-sequencing data are corrected by matching mutual nearest neighbors. Nat. Biotechnol. 36, 421–427 (2018). source
Not checked in the free run.
- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
- Findingcitation
Milo also recovered over-expression of genes involved in regulating leukocyte recruitment, confirming the validated immunomodulatory phenotype of scar-associated tissue.
“We also recovered over-expression of genes associated with regulation of leukocyte recruitment, confirming the validated immunomodulatory phenotype displayed by scar-associated tissue (Supp Fig 14A) [24].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [24] Wilkinson, A. L., Qurashi, M. & Shetty, S. The role of sinusoidal endothelial cells in the axis of inflammation and cancer within the liver. Front. Physiol. 11, 990 (2020). source
Not checked in the free run.
- Findingown data
Applying batch correction in silico improves the performance of differential abundance methods when batch effects are present.
“In silico batch correction enhances the performance of DA methods in the presence of batch effects: comparison of performance of DA methods with no batch effect, with batch effects of increasing magnitude corrected with MNN, and uncorrected batch effects.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
- Findingown data
Milo was the only method that kept false discovery rate controlled as batch effect magnitude increased.
“We found that Milo was the only method to maintain FDR control across batch effects of increasing magnitude (Fig 2E).”
Quote in the paperSubstantial claim- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
- Rests on an assumption: No batch correction algorithm is perfect, and uncorrected or incompletely corrected batch effects can cause false discoveries in differential abundance analysis.
- Findingown data
Milo found a gradient of neighbourhood abundance changes across the endothelial compartment, suggesting a continuous transition between healthy and diseased cell states.
“Milo identified a gradient of changes in neighbourhood abundance across this compartment, suggestive of a continuous transition between healthy and diseased cell states (Fig 5E).”
Quote in the paperSubstantial claim - Findingown data
Cirrhotic endothelium showed down-regulation of genes involved in response to infection, endocytosis and immune complex clearance, including FCN2 and FCN3.
“In addition, cirrhotic endothelium displayed a down-regulation of genes involved in response to infection, endocytosis and immune complex clearance, including FCN2 and FCN3 (Supp Fig 14B)”
Quote in the paperNot a substantial claim - Findingown data
Milo's memory requirements fit within the resources of ordinary desktop computers, under 16 GB.
“These memory requirements are within the resources of common desktop computers (i.e. <16GB).”
Quote in the paperSubstantial claim - Findingown data
Because Milo models sample-to-sample variability, adding more replicates improves the accuracy of its effect estimates.
“Notably, because Milo models sample-to-sample variability, increasing the number of replicates increases the accuracy of the effect estimates (Supp Fig 8).”
Quote in the paperNot a substantial claim - Findingown data
In the mouse thymus analysis, 407 genes were differentially expressed at a 10% false discovery rate.
“Gene Ontology Biological Process term analysis was performed on the 407 DE genes (FDR 10%) using the R package enrichR [47].”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
- Findingown data
Neighbourhoods tend to be homogeneous, with an average of 80% of cells sharing the most common cell type label.
“Neighbourhoods are generally homogeneous, with an average of 80% of cells belonging to the most abundant cell type label.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
- Interpretationreasoning
Milo addresses the problem of finding sets of cells that differ in abundance between conditions without depending on pre-existing clusters.
“Milo tackles a key problem: determining sets of cells that are differentially abundant between conditions without relying on pre-existing sets of clusters.”
Quote in the paperSubstantial claim - Premisecitation
Other clustering-free methods do not use the replication structure of the experimental design to account for technical variability between samples.
“However, these methods do not exploit the replication structure in the experimental design to account for technical variability between samples.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [7] Zhao, J. et al. Detection of differentially abundant cell subpopulations discriminates biological states in scRNA-seq data. Proc. Natl Acad. Sci. USA 118, e2100293118 (2021). source
Not checked in the free run.
- cites [8] Burkhardt, D. B. et al. Quantifying the effect of experimental perturbations at single-cell resolution. Nat. Biotechnol. 39, 619–629 (2021). source
Not checked in the free run.
- Interpretationreasoning
These opposing patterns, seen for example in the T cell (2) and endothelial (5) compartments, were hidden in the earlier study because it relied on pre-clustering.
“These patterns, exemplified by the T cell (2) and the endothelial (5) compartments were obscured in the previous study due to the reliance on pre-clustering (Fig 5D).”
Quote in the paperSubstantial claim - Interpretationown data
Milo can perform differential abundance analysis on large and complex datasets at a scale and speed that is practical on a desktop computer.
“This benchmarking analysis demonstrates that Milo is able to perform differential abundance analysis in large and complex datasets at a scale and speed that is feasible on a desktop computer.”
Quote in the paperSubstantial claim - Study design and conductcitation
Milo handles the non-independence of overlapping neighbourhoods by using a weighted Benjamini-Hochberg procedure in which p-values are weighted by the inverse of neighbourhood connectivity.
“We account for the non-independence of overlapping neighbourhoods by applying a weighted version of the Benjamini-Hochberg (BH) method, where p-values are weighted by the reciprocal of the neighbourhood connectivity”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [6] Lun, A. T. L., Richard, A. C. & Marioni, J. C. Testing for differential abundance in mass cytometry data. Nat. Methods 14, 707–709 (2017). source
Not checked in the free run.
- Interpretationreasoning
Milo uses generalized linear models to support complex experimental designs.
“Our method, which we call Milo, leverages the flexibility of generalized linear models (GLM), thus allowing complex experimental designs”
Quote in the paperSubstantial claim
- May combine two claims. 29
- Classified as probably not a substantial claim. 12
May combine two claims.
Findingown dataMilo detected simulated differentially abundant regions with high sensitivity and kept false discovery rate controlled across benchmarking scenarios.
“Milo detected the simulated DA regions with high sensitivity and maintained FDR control across benchmarking scenarios (Fig 2C, Supp Fig 4).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataMilo's run time increased linearly with dataset size, and a dataset of 200,000 cells took less than 60 minutes.
“Notably, the amount of time taken increased linearly with the total size of the data set (Fig 3A), which for a large set of 200k cells was less than 60 minutes.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataEven when the experimental variable and batch are partially confounded, Milo can still detect differential abundance and control false discovery rate, particularly for large effect sizes despite relatively strong batch effects.
“Even in the presence of partial confounding between the experimental variable and batch, Milo is still able to identify differential abundance and control FDR, particularly for large effect sizes despite relatively strong batch effects (Supp Fig 10).”
Quote in the paperSubstantial claim- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
- Rests on an assumption: No batch correction algorithm is perfect, and uncorrected or incompletely corrected batch effects can cause false discoveries in differential abundance analysis.
May combine two claims.
Findingown dataWith purely random sampling of neighbourhood indices, the number of neighbourhoods needed to cover a fixed proportion of cells grows linearly with the total number of index cells, which raises the multiple testing burden and can lower statistical power.
“Adopting a purely random sampling approach means that the number of neighbourhoods required to sample a fixed proportion of cells scales linearly with the total number of index cells (Supp Fig 1A). This leads to an increased multiple testing burden, with the potential to reduce statistical power.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataIn the human liver dataset, Milo identified 2677 neighbourhoods, of which 1351 showed evidence of differential abundance at a 10% false discovery rate.
“we applied Milo, identifying 2677 neighbourhoods spanning the KNN graph, of which 1351 showed evidence of differential abundance (10% FDR; Fig 5C).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataCells transcriptionally similar to the mTEC-biased Intertypical TEC, called Neighbourhood Group 5, were depleted in older mice, giving independent validation in a dataset of about 90,000 single cells.
“This analysis revealed that the cells transcriptionally similar to the mTEC-biased Intertypical TEC (‘Neighbourhood Group 5’) were indeed depleted in older mice, providing independent validation of our findings in a dataset comprising ~90,000 single-cells (Fig 4I).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataMilo reduces these compositional biases by combining TMM normalisation with a graph spatial false discovery rate, achieving a median false discovery rate of about 10% in those simulations.
“Notably, Milo mitigates against these compositional biases using a combination of TMM normalisation and a graph spatial FDR, which was demonstrated by a median FDR ~10% across these specific simulations (Supp Fig 4A).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataAt a 10% false discovery rate, Milo identified 208 differentially abundant neighbourhoods in mouse thymic epithelial cells, 101 decreasing and 107 increasing with age, spanning multiple TEC states.
“At a 10% FDR, we identified 208 DA neighbourhoods (101 showed a decreased abundance with age, 107 an increased abundance with age) spanning multiple TEC states (Fig 4C).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataCells from younger mice in the depleted Intertypical TEC sub-state showed up-regulation of several cytokine response genes, illustrated by enrichment of the Gene Ontology term response to cytokine with adjusted p-value 0.047.
“This analysis indicated that the cells from younger mice up-regulated multiple cytokine response genes (e.g. Stat1, Stat4, Aff3), illustrated by the enriched Gene Ontology term GO:0034097 ‘response to cytokine’ (enrichment adjusted p-value=0.047).”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataMilo's effect size estimates were unbiased and more accurate for true differentially abundant regions, especially at higher fold changes.
“By contrast, the effect size estimates from Milo were unbiased and were more accurate for true DA regions, especially for higher fold changes.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataMilo recovered the previously reported age-related accumulation of Intertypical TEC and also found an additional subset of these cells that decreased with age.
“Moreover, whilst we recovered the previously reported accumulation of Intertypical TEC with age, we also identified an additional subset of these cells that were depleted with age (Fig 4C-D).”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataIn the mouse thymus dataset, refined sampling with an initial random sample of 30% of cells yielded 363 neighbourhoods.
“The refined sampling, using an initial random sampling of 30% of cells, identified 363 neighbourhoods.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
Probably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataIn the liver cirrhosis dataset, refined sampling with K=30 and an initial sampled cell proportion of 0.05 yielded 2676 neighbourhoods.
“Refined sampling identified 2676 neighbourhoods (K=30, initial proportion of sampled cells = 0.05).”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
May combine two claims.
PremisereasoningPutting single cells into discrete clusters can be problematic, particularly for continuous differentiation, developmental or stimulation trajectories, and this limits the power and resolution of differential abundance testing.
“However, assigning single-cells to discrete clusters can be problematic, especially in the context of continuous differentiation, developmental or stimulation trajectories, thus limiting the power and resolution of such differential abundance (DA) testing strategies.”
Quote in the paperSubstantial claimMay combine two claims.
PremisecitationMELD and DAseq are built for pairwise comparisons and cannot easily be extended to complex experimental designs with continuous variables, multifactorial conditions or nuisance covariates.
“In addition, MELD and DAseq are designed for pairwise comparisons between two biological conditions, and cannot be easily extended to detect differential abundance with complex experimental designs, including continuous variables (age, time points), multifactorial conditions or nuisance covariates.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [7] Zhao, J. et al. Detection of differentially abundant cell subpopulations discriminates biological states in scRNA-seq data. Proc. Natl Acad. Sci. USA 118, e2100293118 (2021). source
Not checked in the free run.
- cites [8] Burkhardt, D. B. et al. Quantifying the effect of experimental perturbations at single-cell resolution. Nat. Biotechnol. 39, 619–629 (2021). source
Not checked in the free run.
May combine two claims.
Interpretationown dataMilo's sensitivity is demonstrated by its identification of an mTEC progenitor state that is depleted with age, a result that clustering approaches did not resolve.
“In summary, these analyses demonstrate the sensitivity of Milo by identifying that a mTEC progenitor state is depleted with age, a finding that was not resolved using clustering approaches.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
InterpretationreasoningRare cell states may be represented by only a few neighbourhoods, which makes identifying marker genes difficult.
“One challenge of this approach is that rare cell states may be represented by a small subset of neighbourhoods, thus making identification of marker genes challenging.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationown dataBy modelling cell counts with a negative binomial GLM, Milo can handle arbitrarily complex experimental designs, as shown by its application to the ageing mouse thymus.
“By modelling cell counts with a NB GLM, Milo can incorporate arbitrarily complex experimental designs as demonstrated by our application of Milo to detect compositional changes in the ageing mouse thymus (Fig 4).”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationnone givenIn experimental design, it is important to avoid complete confounding between technical sources of variation and the experimental variables of interest, including the number of cells collected per condition.
“Moreover, when considering experimental design, it is vital to avoid complete confounding between technical sources of variation and experimental variables of interest, including the number of cells acquired for each condition.”
Quote in the paperNot a substantial claim- Rests on an assumption: Biological sources of variation can confound analyses, for example when age, gender and other biological factors co-vary with the experimental variable of interest.
Probably not a substantial claim: may not pass S3 (load-bearing).
InterpretationcitationMilo may also be useful for single-cell assays beyond scRNA-seq, including multi-omic assays.
“Milo may also be applicable to single-cell assays other than scRNA-seq, including multi-omic assays [35–39].”
Quote in the paperNot a substantial claim0 of 5 sources checked- cites [35] Stoeckius, M. et al. Simultaneous epitope and transcriptome measurement in single cells. Nat. Methods 14, 865–868 (2017). source
Not checked in the free run.
- cites [36] Cao, J. et al. Joint profiling of chromatin accessibility and gene expression in thousands of single cells. Science 361, 1380–1385 (2018). source
Not checked in the free run.
- cites [37] Chen, S., Lake, B. B. & Zhang, K. High-throughput sequencing of the transcriptome and chromatin accessibility in the same cell. Nat. Biotechnol. 37, 1452–1457 (2019). source
Not checked in the free run.
- cites [38] Zhu, C. et al. An ultra high-throughput method for single-cell joint analysis of open chromatin and transcriptome. Nat. Struct. Mol. Biol. 26, 1063–1070 (2019). source
Not checked in the free run.
- cites [39] Ma, S. et al. Chromatin potential identified by shared single-cell profiling of RNA and chromatin. Cell 183, 1103–1116 (2020). source
Not checked in the free run.
May combine two claims.
Interpretationown dataMilo's flexibility lets it account for batch effects of varying magnitude and control false discoveries in a way other differential abundance algorithms cannot.
“This flexibility is illustrated in the ability of Milo to account appropriately for batch effects of varying magnitude, controlling false discoveries that is not possible using other DA testing algorithms (Fig 2E).”
Quote in the paperSubstantial claim- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
May combine two claims.
InterpretationcitationBecause cytokine signalling is important for mTEC differentiation, these younger-mouse TEC may be differentiating more efficiently toward the mTEC lineage.
“Cytokine signaling is key to mTEC differentiation [22,23], indicating that these TEC from younger mice might be differentiating more efficiently to the mTEC lineage.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [22] Akiyama, T. et al. The tumor necrosis factor family receptors RANK and CD40 cooperatively establish the thymic medullary microenvironment and self-tolerance. Immunity 29, 423–437 (2008). source
Not checked in the free run.
- cites [23] Hikosaka, Y. et al. The cytokine RANKL produced by positively selected thymocytes fosters medullary thymic epithelial cells that express autoimmune regulator. Immunity 29, 438–450 (2008). source
Not checked in the free run.
May combine two claims.
Interpretationown dataAcross a range of experimental scenarios, including residual batch effects, Milo outperforms alternative differential abundance testing methods.
“In summary, we show that Milo outperforms alternative methods for differential abundance testing across a range of experimental scenarios, including in the presence of residual batch effects.”
Quote in the paperSubstantial claim- Rests on an assumption: Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
May combine two claims.
Interpretationown dataBatch integration before graph building and inclusion of known confounders in the testing framework can reduce nuisance effects, but these approaches may lose biological signal when the condition of interest and confounders are strongly correlated.
“While we have shown that nuisance effects can be minimized by applying batch integration prior to graph building (Supp Fig 9) and by incorporating known confounders in the testing framework (Fig 2E, Supp Fig 11), these strategies can lead to loss of biological signal if the condition of interest and the confounders are strongly correlated.”
Quote in the paperSubstantial claim- Rests on an assumption: Biological sources of variation can confound analyses, for example when age, gender and other biological factors co-vary with the experimental variable of interest.
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).
InterpretationreasoningMilo could help discover fundamental biological and clinically relevant processes across multiple layers of molecular regulation when these are measured at single-cell resolution.
“Thus, Milo has the potential to facilitate the discovery of fundamental biological and clinically relevant processes across multiple layers of molecular regulation when they are assayed at single-cell resolution.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationExisting clustering-free differential abundance methods either fail to model variability in cell numbers between replicate samples or are mainly built for pairwise comparisons, which restricts their use in more complex experimental designs.
“However, these have different limitations, where they either do not model variability in cell numbers between replicate samples or are primarily designed for pairwise comparisons. This limits their application to datasets with more complex experimental designs”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [6] Lun, A. T. L., Richard, A. C. & Marioni, J. C. Testing for differential abundance in mass cytometry data. Nat. Methods 14, 707–709 (2017). source
Not checked in the free run.
- cites [7] Zhao, J. et al. Detection of differentially abundant cell subpopulations discriminates biological states in scRNA-seq data. Proc. Natl Acad. Sci. USA 118, e2100293118 (2021). source
Not checked in the free run.
- cites [8] Burkhardt, D. B. et al. Quantifying the effect of experimental perturbations at single-cell resolution. Nat. Biotechnol. 39, 619–629 (2021). source
Not checked in the free run.
May combine two claims.
Interpretationnone givenCells in one neighbourhood do not necessarily form a unique biological subpopulation, because a cellular state can span several neighbourhoods.
“Secondly, cells in a single neighbourhood do not necessarily represent a unique biological subpopulation; a cellular state might span multiple neighbourhoods.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationnone givenModelling cell states as overlapping neighbourhoods lets Milo locate perturbed cellular states accurately and identify the molecular programs behind them.
“Moreover, by modelling cell states as overlapping neighbourhoods, we are able to accurately pinpoint the perturbed cellular states, enabling the identification of the underlying molecular programs.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Interpretationnone givenReliable differential abundance results require well-designed single-cell experiments, and biological replicates are needed to estimate the negative binomial dispersion parameter for each neighbourhood.
“First, reliable results from DA testing depend on well-designed single-cell experiments. Biological replicates are required to estimate the negative binomial dispersion parameter for each neighbourhood.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationown dataMilo can be applied to many datasets with different topologies, including gradual state transitions, which removes the need for iterative sub-clustering and reveals subtle differential abundance differences that would otherwise be hidden.
“A strength of our approach is that it is applicable to a wide range of datasets with different topologies, including gradual state transitions, thus removing the need for time-consuming iterative sub-clustering, and identifying subtle differences in differential abundance that would otherwise be obscured (Fig 5D).”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.
InterpretationreasoningThere are situations where differential abundance analysis on clusters or predefined cell populations is preferable, such as when differences are visible in large clusters or when comparing pre-annotated populations across datasets without integrating single cells on a common manifold.
“Thirdly, there will be cases where DA analysis on clusters or predefined cell populations will be preferable, for example where differences are apparent in large clusters by visualization, or to compare abundances of pre-annotated cell populations across datasets without requiring integration of single-cells on a common manifold.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningMilo's definition of neighbourhoods overcomes the main limitations of standard clustering-based differential abundance analysis while using graphs, a common data structure in single-cell analysis.
“The definition of neighbourhoods, as implemented in Milo, overcomes the main limitations of standard-of-practice clustering-based DA analysis, whilst using a common data-structure in single-cell analysis - graphs.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
InterpretationcitationMilo works with established single-cell analysis workflows and is available as an open-source R software package.
“Moreover, Milo is fully interoperable with established single-cell analysis workflows and is implemented as an open-source R software package [30]”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [30] R Core Team. R: A Language and Environment for Statistical Computing. https://www.R-project.org (R Foundation for Statistical Computing, 2017).
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationDAseq assigns each cell a differential abundance score based on the relative prevalence of cells from the two biological states in its neighbourhood, using a range of K values.
“DAseq computes for each cell a DA score based on the relative prevalence of cells from both biological states in the cell’s neighborhood, using a range of K values.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [7] Zhao, J. et al. Detection of differentially abundant cell subpopulations discriminates biological states in scRNA-seq data. Proc. Natl Acad. Sci. USA 118, e2100293118 (2021). source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationMELD estimates the probability density of each sample across a KNN graph and uses this to quantify the relative likelihood that a cell would be observed in one condition versus the others.
“MELD estimates the probability density of each sample over a KNN graph, which is then used to quantify the relative likelihood that each cell would be observed in one condition relative to the others.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [8] Burkhardt, D. B. et al. Quantifying the effect of experimental perturbations at single-cell resolution. Nat. Biotechnol. 39, 619–629 (2021). source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
PremisecitationMELD does not carry out statistical inference; instead, the user sets a threshold on per-cell likelihoods to classify a cell as belonging to a differentially abundant region or not.
“MELD does not perform any statistical inference, instead the user selects a threshold on the per-cell likelihoods to define a cell as being in a DA region or not.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [8] Burkhardt, D. B. et al. Quantifying the effect of experimental perturbations at single-cell resolution. Nat. Biotechnol. 39, 619–629 (2021). source
Not checked in the free run.
The paper relies on this without support.
The gene signatures associated with the ductular reaction in human cirrhosis are largely unexplored.
“However, the gene signatures associated with this process in human cirrhosis are largely unexplored.”
The paper relies on this without support.
Technical batch effects can pose a particular challenge to differential abundance testing methods, including all methods benchmarked in the paper.
“Technical batch effects have the potential to pose a particular challenge to differential abundance testing methods, including all of those benchmarked above.”
The paper relies on this without support.
Biological sources of variation can confound analyses, for example when age, gender and other biological factors co-vary with the experimental variable of interest.
“Moreover, biological sources of variation may confound analyses, such as the co-variation of age, gender and other biological factors with the experimental variable of interest.”
The paper relies on this without support.
No batch correction algorithm is perfect, and uncorrected or incompletely corrected batch effects can cause false discoveries in differential abundance analysis.
“However, no batch correction algorithm is perfect, and uncorrected or incompletely corrected batch effects may lead to false discoveries in differential abundance analysis.”
The authors state this as an assumption.
The KNN graph is a faithful representation of single-cell phenotypes, where cell-cell similarities are driven by true biological effects rather than technical or batch effects.
“We assume that the KNN graph is a faithful representation of the single cell phenotypes, where cell-cell similarities are driven by true biological effects, rather than technical/batch effects.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Several distance measures for scRNA-seq have been explored, including Wasserstein distance, maximum mean discrepancy, neighborhood-based measures, and energy distance.
“Multiple distance measures for scRNA-seq have been explored by the single-cell community in recent years, including Wasserstein distance computed in an optimal transport framework22,23, maximum mean discrepancy24, neighborhood-based measures17,18, and energy distance (E-distance)20.”
From scPerturb: harmonized single-cell perturbation data (2024) · cites this paper as [18]
Studies with mixtures of cell types have developed complex methods for quantifying similarity between heterogeneous cell populations.
“Studies with mixtures of cell types have developed complex methods for quantifying similarity between heterogeneous cell populations17-19.”
From scPerturb: harmonized single-cell perturbation data (2024) · cites this paper as [18]
Milo is a method to detect compositional changes occurring in smaller subpopulations of cells, designated as neighbourhoods over the KNN graph of cell-cell similarities.
“Milo[34] is a method to detect compositional changes occurring in smaller subpopulations of cells, designated as neighbourhoods over the k-nearest neighbour (KNN) graph of cell-cell similarities.”
From A mini-review on perturbation modelling across single-cell omic modalities (2024) · cites this paper as [34]
Milo accurately identifies emerging perturbed cellular states without the need of a priori cell annotation, facilitating identification of underlying disease-related molecular motifs.
“By representing cell states as overlapping neighbourhoods, Milo accurately identifies emerging perturbed cellular states (without the need of a priori cell annotation), facilitating the identification of underlying disease-related molecular motifs.”
From A mini-review on perturbation modelling across single-cell omic modalities (2024) · cites this paper as [34]
Reducing K for the KNN graph or constructing a graph on particular lineages mitigates the loss of information in KNN-based methods.
“Reducing K for the KNN graph or constructing a graph on particular lineages mitigates this issue112.”
From Best practices for single-cell analysis across modalities (2023) · cites this paper as [112]
KNN-based methods are sensitive to a loss of information if the conditions of interest and confounding sources of variation are strongly correlated.
“KNN-based methods are sensitive to a loss of information if the conditions of interest and confounding sources of variation are strongly correlated.”
From Best practices for single-cell analysis across modalities (2023) · cites this paper as [112]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Liver neighbourhood count differs between sections
MinorThe Results report 2,677 neighbourhoods for the liver analysis, while the Methods report 2,676 for the same analysis.
“we applied Milo, identifying 2677 neighbourhoods spanning the KNN graph”
“Refined sampling identified 2676 neighbourhoods (K=30, initial proportion of sampled cells = 0.05).”
Cites
Robinson, M. D., McCarthy, D. J., & Smyth, G. K. (2010) · Bioinformatics · doi:10.1093/bioinformatics/btp616 · Cited by 41,364 · see citing works
limma powers differential expression analyses for RNA-sequencing and microarray studies
Ritchie, M. E., Phipson, B., Wu, D., Hu, Y., Law, C. W., Shi, W., & Smyth, G. K. (2015) · Nucleic Acids Research · doi:10.1093/nar/gkv007 · Cited by 40,458 · see citing works
clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters
Yu, G., Wang, L.-G., Han, Y., & He, Q.-Y. (2012) · OMICS: A Journal of Integrative Biology · doi:10.1089/omi.2011.0118 · Cited by 36,259 · see citing works
Fast unfolding of communities in large networks
Blondel, V. D., Guillaume, J.-L., Lambiotte, R., & Lefebvre, E. (2008) · Journal of Statistical Mechanics: Theory and Experiment · doi:10.1088/1742-5468/2008/10/p10008 · Cited by 21,424 · see citing works
Bioconductor: open software development for computational biology and bioinformatics
Gentleman, R., Carey, V., Bates, D., Bolstad, B., Dettling, M., Dudoit, S., Ellis, B., Gautier, L., Ge, Y., Gentry, J., et al. (2004) · Genome Biology · doi:10.1186/gb-2004-5-10-r80 · Cited by 13,133 · see citing works
Enrichr: a comprehensive gene set enrichment analysis web server 2016 update
Kuleshov, M. V., Jones, M. R., Rouillard, A. D., Fernandez, N. F., Duan, Q., Wang, Z., Koplev, S., Jenkins, S. L., Jagodnik, K. M., Lachmann, A., et al. (2016) · Nucleic Acids Research · doi:10.1093/nar/gkw377 · Cited by 10,799 · see citing works
The igraph software package for complex network research
Csárdi, G., & Nepusz, T. (2006) · Medical Entomology and Zoology · Cited by 10,078 · see citing works
A scaling normalization method for differential expression analysis of RNA-seq data
Robinson, M. D., & Oshlack, A. (2010) · Genome Biology · doi:10.1186/gb-2010-11-3-r25 · Cited by 7,876 · see citing works
Comparison of the predicted and observed secondary structure of T4 phage lysozyme
Matthews, B. W. (1975) · Biochimica et Biophysica Acta (BBA) - Protein Structure · doi:10.1016/0005-2795(75)90109-9 · Cited by 5,399 · see citing works
McCarthy, D. J., Chen, Y., & Smyth, G. K. (2012) · Nucleic Acids Research · doi:10.1093/nar/gks042 · Cited by 5,195 · see citing works
Orchestrating high-throughput genomic analysis with Bioconductor
Huber, W. K., Carey, V. J., Gentleman, R. C., Anders, S., Carlson, M. R. J., Carvalho, B. S., Bravo, H. C., Davis, S. R., Gatto, L., Girke, T., et al. (2015) · Nature Methods · doi:10.1038/nmeth.3252 · Cited by 3,499 · see citing works
Self-citationSimultaneous epitope and transcriptome measurement in single cells
Stoeckius, M., Hafemeister, C., Stephenson, W., Houck-Loomis, B., Chattopadhyay, P. K., Swerdlow, H., Satija, R., & Smibert, P. (2017) · Nature Methods · doi:10.1038/nmeth.4380 · Cited by 3,098 · see citing works
Batch effects in single-cell RNA-sequencing data are corrected by matching mutual nearest neighbors
Haghverdi, L., Lun, A. T. L., Morgan, M. D., & Marioni, J. C. (2018) · Nature Biotechnology · doi:10.1038/nbt.4091 · Cited by 2,416 · see citing works
Self-citationCurrent best practices in single-cell RNA-seq analysis: a tutorial
Luecken, M. D., & Theis, F. J. (2019) · Molecular Systems Biology · doi:10.15252/msb.20188746 · Cited by 2,231 · see citing works
Computing Communities in Large Networks Using Random Walks
Pons, P., & Matthieu Latapy, B. (2005) · Lecture notes in computer science · doi:10.1007/11569596_31 · Cited by 2,147 · see citing works
Dixit, A., Parnas, O., Li, B., Chen, J., Fulco, C. P., Jerby-Arnon, L., Marjanovic, N. D., Dionne, D., Burks, T., Raychowdhury, R., et al. (2016) · Cell · doi:10.1016/j.cell.2016.11.038 · Cited by 1,863 · see citing works
Resolving the fibrotic niche of human liver cirrhosis at single-cell level
Ramachandran, P., Dobie, R., Wilson-Kanamori, J. R., Dora, E. F., Henderson, B. E. P., Luu, N. T., Portman, J. R., Matchett, K. P., Brice, M., Marwick, J. A., et al. (2019) · Nature · doi:10.1038/s41586-019-1631-3 · Cited by 1,784 · see citing works
Self-citationA single-cell survey of the small intestinal epithelium
Haber, A. L., Biton, M., Rogel, N., Herbst, R. H., Shekhar, K., Smillie, C., Burgin, G., Delorey, T. M., Howitt, M. R., Katz, Y., et al. (2017) · Nature · doi:10.1038/nature24489 · Cited by 1,750 · see citing works
A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor
Lun, A. T. L., McCarthy, D. J., & Marioni, J. C. (2016) · F1000Research · doi:10.12688/f1000research.9501.2 · Cited by 1,586 · see citing works
Self-citationA step-by-step workflow for low-level analysis of single-cell RNA-seq data
Lun, A. T. L., McCarthy, D. J., & Marioni, J. C. (2016) · F1000Research · doi:10.12688/f1000research.9501.1 · Cited by 1,511 · see citing works
Self-citationBenchmarking atlas-level data integration in single-cell genomics
Luecken, M. D., Büttner, M., Chaichoompu, K., Danese, A., Interlandi, M., Mueller, M. F., Strobl, D. C., Zappia, L., Dugas, M., Colomé-Tatché, M., et al. (2022) · Nature Methods · doi:10.1038/s41592-021-01336-8 · Cited by 1,438 · see citing works
Challenges in unsupervised clustering of single-cell RNA-seq data
Kiselev, V. Y., Andrews, T. S., & Hemberg, M. (2019) · Nature Reviews Genetics · doi:10.1038/s41576-018-0088-9 · Cited by 1,149 · see citing works
Chromatin Potential Identified by Shared Single-Cell Profiling of RNA and Chromatin
Ma, S., Zhang, B., LaFave, L. M., Earl, A. S., Chiang, Z., Hu, Y., Ding, J., Brack, A., Kartha, V. K., Tay, T., et al. (2020) · Cell · doi:10.1016/j.cell.2020.09.056 · Cited by 1,146 · see citing works
A benchmark of batch-effect correction methods for single-cell RNA sequencing data
Tran, H. T. N., Ang, K. S., Chevrier, M., Zhang, X., Lee, N. Y. S., Goh, M., & Chen, J. (2020) · Genome Biology · doi:10.1186/s13059-019-1850-9 · Cited by 1,113 · see citing works
Pooled CRISPR screening with single-cell transcriptome readout.
Datlinger, P., Rendeiro, A. F., Schmidl, C., Krausgruber, T., Traxler, P., Klughammer, J., Schuster, L. C., Kuchler, A., Alpar, D., & Bock, C. (2017) · Nature Methods · doi:10.1038/nmeth.4177 · Cited by 1,096 · see citing works
Stoeckius, M., Zheng, S., Houck-Loomis, B., Hao, S., Yeung, B. Z., Mauck, W. M., Smibert, P., & Satija, R. (2018) · Genome Biology · doi:10.1186/s13059-018-1603-1 · Cited by 1,046 · see citing works
A single-cell molecular map of mouse gastrulation and early organogenesis
Pijuan-Sala, B., Griffiths, J. A., Guibentif, C., Hiscock, T. W., Jawaid, W., Calero-Nieto, F. J., Mulas, C., Ibarra-Soria, X., Tyser, R. C. V., Ho, D. L. L., et al. (2019) · Nature · doi:10.1038/s41586-019-0933-9 · Cited by 1,033 · see citing works
Self-citationOrchestrating single-cell analysis with Bioconductor
Amezquita, R. A., Lun, A. T. L., Becht, E., Carey, V. J., Carpp, L. N., Geistlinger, L., Marini, F., Rue-Albrecht, K., Risso, D., Soneson, C., et al. (2020) · Nature Methods · doi:10.1038/s41592-019-0654-x · Cited by 945 · see citing works
Self-citationJoint profiling of chromatin accessibility and gene expression in thousands of single cells
Cao, J., Cusanovich, D. A., Ramani, V., Aghamirzaie, D., Pliner, H. A., Hill, A. J., Daza, R. M., McFaline-Figueroa, J. L., Packer, J. S., Christiansen, L., et al. (2018) · Science · doi:10.1126/science.aau0730 · Cited by 938 · see citing works
High-throughput sequencing of the transcriptome and chromatin accessibility in the same cell
Chen, S., Lake, B. B., & Zhang, K. (2019) · Nature Biotechnology · doi:10.1038/s41587-019-0290-0 · Cited by 810 · see citing works
Dissecting Immune Circuits by Linking CRISPR-Pooled Screens with Single-Cell RNA-Seq
Jaitin, D. A., Weiner, A., Yofe, I., Lara-Astiaso, D., Keren-Shaul, H., David, E., Salame, T. M., Tanay, A., van Oudenaarden, A., & Amit, I. (2016) · Cell · doi:10.1016/j.cell.2016.11.039 · Cited by 777 · see citing works
Bias, robustness and scalability in single-cell differential expression analysis
Soneson, C., & Robinson, M. D. (2018) · Nature Methods · doi:10.1038/nmeth.4612 · Cited by 770 · see citing works
MULTI-seq: sample multiplexing for single-cell RNA sequencing using lipid-tagged indices
McGinnis, C. S., Patterson, D. M., Winkler, J., Conrad, D. N., Hein, M. Y., Srivastava, V., Hu, J. L., Murrow, L. M., Weissman, J. S., Werb, Z., et al. (2019) · Nature Methods · doi:10.1038/s41592-019-0433-8 · Cited by 643 · see citing works
Wishbone identifies bifurcating developmental trajectories from single-cell data
Setty, M., Tadmor, M. D., Reich-Zeliger, S., Angel, O., Salame, T. M., Kathail, P., Choi, K., Bendall, S., Friedman, N., & Pe'er, D. (2016) · Nature Biotechnology · doi:10.1038/nbt.3569 · Cited by 558 · see citing works
Akiyama, T., Shimo, Y., Yanai, H., Qin, J., Ohshima, D., Maruyama, Y., Asaumi, Y., Kitazawa, J., Takayanagi, H., Penninger, J. M., et al. (2008) · Immunity · doi:10.1016/j.immuni.2008.06.015 · Cited by 503 · see citing works
Hikosaka, Y., Nitta, T., Ohigashi, I., Yano, K., Ishimaru, N., Hayashi, Y., Matsumoto, M., Matsuo, K., Penninger, J. M., Takayanagi, H., et al. (2008) · Immunity · doi:10.1016/j.immuni.2008.06.018 · Cited by 433 · see citing works
Ductular reaction in liver diseases: pathological mechanisms and translational significances
Sato, K., Marzioni, M., Meng, F., Francis, H., Glaser, S., & Alpini, G. (2019) · Hepatology · doi:10.1002/hep.30150 · Cited by 423 · see citing works
scCODA is a Bayesian model for compositional single-cell data analysis
Büttner, M., Ostner, J., Müller, C. L., Theis, F. J., & Schubert, B. (2021) · Nature Communications · doi:10.1038/s41467-021-27150-6 · Cited by 388 · see citing works
An ultra high-throughput method for single-cell joint analysis of open chromatin and transcriptome
Zhu, C., Yu, M., Huang, H., Juric, I., Abnousi, A., Hu, R., Lucero, J., Behrens, M. M., Hu, M., & Ren, B. (2019) · Nature Structural & Molecular Biology · doi:10.1038/s41594-019-0323-x · Cited by 344 · see citing works
Lund, S. P., Nettleton, D. S., McCarthy, D. J., & Smyth, G. K. (2012) · Statistical Applications in Genetics and Molecular Biology · doi:10.1515/1544-6115.1826 · Cited by 322 · see citing works
Multiple Hypotheses Testing with Weights
Benjamini, Y., & Hochberg, Y. (1997) · Scandinavian Journal of Statistics · doi:10.1111/1467-9469.00072 · Cited by 311 · see citing works
Quantifying the effect of experimental perturbations at single-cell resolution
Burkhardt, D. B., Stanley, J. S., Tong, A., Perdigoto, A. L., Gigante, S. A., Herold, K. C., Wolf, G., Giraldez, A. J., van Dijk, D., & Krishnaswamy, S. (2021) · Nature Biotechnology · doi:10.1038/s41587-020-00803-5 · Cited by 226 · see citing works
Ageing compromises mouse thymus function and remodels epithelial cell differentiation
Baran‐Gale, J., Morgan, M. D., Maio, S., Dhalla, F., Calvo‐Asensio, I., Deadman, M. E., Handel, A. E., Maynard, A., Chen, S., Green, F., et al. (2020) · eLife · doi:10.7554/elife.56221 · Cited by 182 · see citing works
Self-citationTesting for differential abundance in mass cytometry data
Lun, A. T. L., Richard, A. C., & Marioni, J. C. (2017) · Nature Methods · doi:10.1038/nmeth.4295 · Cited by 181 · see citing works
Self-citationWhole-exome sequencing for finding de novo mutations in sporadic mental retardation
Robinson, P. (2010) · Genome Biology · doi:10.1186/gb-2010-11-12-144 · Cited by 180 · see citing works
FcγRIIb on Liver Sinusoidal Endothelium Clears Small Immune Complexes
Ganesan, L. P., Kim, J., Wu, Y., Mohanty, S., Phillips, G. S., Birmingham, D. J., Robinson, J. M., & Anderson, C. L. (2012) · The Journal of Immunology · doi:10.4049/jimmunol.1202017 · Cited by 166 · see citing works
Detection of differentially abundant cell subpopulations in scRNA-seq data
Zhao, J., Jaffe, A., Li, H., Lindenbaum, O., Sefik, E., Jackson, R., Cheng, X., Flavell, R. A., & Kluger, Y. (2021) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2100293118 · Cited by 154 · see citing works
Spearheading future omics analyses using dyngen, a multi-modal simulator of single cells
Cannoodt, R., Saelens, W., Deconinck, L., & Saeys, Y. (2021) · Nature Communications · doi:10.1038/s41467-021-24152-2 · Cited by 124 · see citing works
dyngen: a multi-modal simulator for spearheading new single-cell omics analyses
Cannoodt, R., Saelens, W., Deconinck, L., & Saeys, Y. (2020) · doi:10.1101/2020.02.06.936971 · Cited by 124 · see citing works
Benchmarking atlas-level data integration in single-cell genomicsPreprint
Luecken, M. D., Büttner, M., Chaichoompu, K., et al. (2020) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2020.05.22.111161 · Cited by 122 · see citing works
The Role of Sinusoidal Endothelial Cells in the Axis of Inflammation and Cancer Within the Liver
Wilkinson, A. L., Qurashi, M., & Shetty, S. (2020) · Frontiers in Physiology · doi:10.3389/fphys.2020.00990 · Cited by 121 · see citing works
Trajectories of cell-cycle progression from fixed cell populations
Gut, G., Tadmor, M. D., Pe’er, D., Pelkmans, L., & Liberali, P. (2015) · Nature Methods · doi:10.1038/nmeth.3545 · Cited by 100 · see citing works
Flexible comparison of batch correction methods for single-cell RNA-seq using BatchBench
Chazarra-Gil, R., van Dongen, S., Kiselev, V. Y., & Hemberg, M. (2021) · Nucleic Acids Research · doi:10.1093/nar/gkab004 · Cited by 93 · see citing works
New insights on the role of vascular endothelial growth factor in biliary pathophysiology
Mariotti, V., Fiorotto, R., Cadamuro, M., Fabris, L., & Strazzabosco, M. (2021) · JHEP Reports · doi:10.1016/j.jhepr.2021.100251 · Cited by 87 · see citing works
Calcium Signaling in Liver Injury and Regeneration
Oliva‐Vilarnau, N., Hankeová, S., Vorrink, S. U., Mkrtchian, S., Andersson, E. R., & Lauschke, V. M. (2018) · Frontiers in Medicine · doi:10.3389/fmed.2018.00192 · Cited by 84 · see citing works
Vascular biology of the biliary epithelium
Morell, C. M., Fabris, L., & Strazzabosco, M. (2013) · Journal of Gastroenterology and Hepatology · doi:10.1111/jgh.12022 · Cited by 38 · see citing works
Carter, L. V., Dooley, D., Elliott, C. E., et al. (1980) · The Medical Journal of Australia · doi:10.5694/j.1326-5377.1980.tb134922.x · Cited by 26 · see citing works
Földi, I., Tornai, T., Tornai, D., et al. (2017) · Liver International · doi:10.1111/liv.13368 · Cited by 26 · see citing works
Calcium Signaling in Cholangiocytes: Methods, Mechanisms, and Effects
Rodrigues, M. Â., Gomes, D. A., & Nathanson, M. H. (2018) · International Journal of Molecular Sciences · doi:10.3390/ijms19123913 · Cited by 19 · see citing works
scCODA: A Bayesian model for compositional single-cell data analysisPreprint
Büttner, M., Ostner, J., Müller, C. L., et al. (2020) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2020.12.14.422688 · Cited by 17 · see citing works
Zhao, J., Jaffe, A., Li, H., Lindenbaum, O., Sefik, E., Jackson, R. P., Cheng, X., Flavell, R. A., & Kluger, Y. (2019) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/711929 · Cited by 8 · see citing works
Expanding immunologic sensitization in children.
Hg, R. (1967) · PubMed · Cited by 4 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Best practices for single-cell analysis across modalities
Heumos, L., Schaar, A. C., Lance, C., Litinetskaya, A., Drost, F., Zappia, L., Lücken, M. D., Strobl, D. C., Henao, J., Curion, F., et al. (2023) · Nature Reviews Genetics · doi:10.1038/s41576-023-00586-w · Cited by 1,051 · see citing works
“3).”
Self-citationQuantifying the effect of experimental perturbations at single-cell resolution
Burkhardt, D. B., Stanley, J. S., Tong, A., Perdigoto, A. L., Gigante, S. A., Herold, K. C., Wolf, G., Giraldez, A. J., van Dijk, D., & Krishnaswamy, S. (2021) · Nature Biotechnology · doi:10.1038/s41587-020-00803-5 · Cited by 226 · see citing works
Cottam, M. A., Caslin, H. L., Winn, N. C., & Hasty, A. H. (2022) · Nature Communications · doi:10.1038/s41467-022-30646-4 · Cited by 216 · see citing works
SEACells infers transcriptional and epigenomic cellular states from single-cell genomics data
Persad, S., Choo, Z.-N., Dien, C., Sohail, N., Masilionis, I., Chaligné, R., Nawy, T., Brown, C. C., Sharma, R., Pe’er, I., et al. (2023) · Nature Biotechnology · doi:10.1038/s41587-023-01716-9 · Cited by 222 · see citing works
A spatially resolved atlas of the human lung characterizes a gland-associated immune niche
Madissoon, E., Oliver, A. J., Kleshchevnikov, V., Wilbrey-Clark, A., Polanski, K., Richoz, N., Ribeiro Orsi, A., Mamanova, L., Bolt, L., Elmentaite, R., et al. (2023) · Nature Genetics · doi:10.1038/s41588-022-01243-4 · Cited by 219 · see citing works
Self-citationscPerturb: harmonized single-cell perturbation data
Peidli, S., Green, T. D., Shen, C., Gross, T., Min, J., Garda, S., Yuan, B., Schumacher, L. J., Taylor-King, J. P., Marks, D. S., et al. (2024) · Nature Methods · doi:10.1038/s41592-023-02144-y · Cited by 194 · see citing works
Zheng, X., Wang, X., Cheng, X., Liu, Z., Yin, Y., Li, X., Huang, Z., Wang, Z., Guo, W., Ginhoux, F., et al. (2023) · Nature Cancer · doi:10.1038/s43018-023-00599-8 · Cited by 172 · see citing works
Barriga, F. M., Tsanov, K. M., Ho, Y.-J., Sohail, N., Zhang, A., Baslan, T., Wuest, A. N., Del Priore, I., Meškauskaitė, B., Livshits, G., et al. (2022) · Nature Cancer · doi:10.1038/s43018-022-00443-5 · Cited by 125 · see citing works
Single-cell transcriptomics for the assessment of cardiac disease
Miranda, A. M. A., Janbandhu, V., Maatz, H., Kanemaru, K., Cranley, J., Teichmann, S. A., Hübner, N., Schneider, M. D., Harvey, R. P., & Noseda, M. (2023) · Nature Reviews Cardiology · doi:10.1038/s41569-022-00805-7 · Cited by 117 · see citing works
Self-citationA prenatal skin atlas reveals immune regulation of human skin morphogenesis
Gopee, N. H., Winheim, E., Olabi, B., Admane, C., Foster, A. R., Huang, N., Botting, R. A., Torabi, F., Sumanaweera, D., Le, A. P., et al. (2024) · Nature · doi:10.1038/s41586-024-08002-x · Cited by 108 · see citing works
Self-citationCharacterizing cellular heterogeneity in chromatin state with scCUT&Tag-pro
Zhang, B., Srivastava, A., Mimitou, E., Stuart, T., Raimondi, I., Hao, Y., Smibert, P., & Satija, R. (2022) · Nature Biotechnology · doi:10.1038/s41587-022-01250-0 · Cited by 104 · see citing works
Causal identification of single-cell experimental perturbation effects with CINEMA-OT
Dong, M., Wang, B., Wei, J., de O. Fonseca, A. H., Perry, C. J., Frey, A., Ouerghi, F., Foxman, E. F., Ishizuka, J. J., Dhodapkar, R. M., et al. (2023) · Nature Methods · doi:10.1038/s41592-023-02040-5 · Cited by 94 · see citing works
Predicting cellular responses to perturbation across diverse contexts with State
Adduri, A. K., Gautam, D., Bevilacqua, B., Naghipourfar, M., Imran, A., Shah, R., Teyssier, N., Verma, R., Carpenter, C., Eraslan, B., et al. (2026) · Cell · doi:10.1016/j.cell.2026.07.052 · Cited by 100 · see citing works
Multi-zonal liver organoids from human pluripotent stem cells
Reza, H. A., Santangelo, C., Iwasawa, K., Reza, A. A., Sekiya, S., Glaser, K., Bondoc, A., Merola, J., & Takebe, T. (2025) · Nature · doi:10.1038/s41586-025-08850-1 · Cited by 71 · see citing works
A mini-review on perturbation modelling across single-cell omic modalities
Gavriilidis, G. I., Vasileiou, V., Orfanou, A., Ishaque, N., & Psomopoulos, F. (2024) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2024.04.058 · Cited by 54 · see citing works
Matatall, K. A., Wathan, T. K., Nguyen, M., Chen, H., McDonald, A., Qi, G., Belk, J. A., Florez, M. A., Le, D. T., Olarinde, T., et al. (2025) · Cell Stem Cell · doi:10.1016/j.stem.2025.06.006 · Cited by 23 · see citing works
StabMap: Mosaic single cell data integration using non-overlapping features
Ghazanfar, S., Guibentif, C., & Marioni, J. C. (2022) · doi:10.1101/2022.02.24.481823 · Cited by 16 · see citing works
Self-citationCellDrift: inferring perturbation responses in temporally sampled single-cell data
Jin, K., Schnell, D., Li, G., Salomonis, N., Prasath, V. B. S., Szczesniak, R., & Aronow, B. J. (2022) · Briefings in Bioinformatics · doi:10.1093/bib/bbac324 · Cited by 11 · see citing works
Robust identification of perturbed cell types in single-cell RNA-seq data
Nicol, P. B., Paulson, D., Qian, G., Liu, X. S., Irizarry, R., & Sahu, A. D. (2023) · doi:10.1101/2023.05.06.539326 · Cited by 33 · see citing works
A comparative transcriptomic analysis of mouse demyelination models and multiple sclerosis lesions
Aboelnour, E. L., Vanoverbeke, V. R., Maupin, E. A., Hatfield, M. M., & Adams, K. L. (2026) · Nature Communications · doi:10.1038/s41467-026-72383-y · Cited by 4 · see citing works
“Milopy, the Python implementation of the Milo toolset, was used to calculate the DA between neighboring cells in the cell type-speci fi c embeddings 23 .”
- Authors
- Dann, E., Henderson, N. C., Teichmann, S. A., Morgan, M. D., & Marioni, J. C.
- Affiliations
- Wellcome Sanger Institute (GB)
- Centre for Inflammation Research (GB)
- University of Cambridge (GB)
- European Bioinformatics Institute (GB)
- Corresponding author
- Morgan, Michael D.
- Marioni, John C.
- Venue
- Nature Biotechnology
- Published
- 2021-09-30
- Volume
- 40
- Issue
- 2
- Pages
- 245-253
- ISSN
- 0733-222X (ISSN-L)
- 1087-0156
- 1546-1696
- PMC
- PMC7617075
- OpenAlex
- W3203496375
- Semantic Scholar
- 8ec438059055824d97a6e0d68ad91a284f526b8c
- Cited by
- 938
- References
- 62
- Citations by year
- 2026: 256 · 2025: 275 · 2024: 192 · 2023: 121 · 2022: 83 · 2021: 11
- FWCI
- 32.81
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Locations
- Nature Biotechnology (publishedVersion)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- Funders
- Wellcome Sanger Institute Core Award
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene expression and cancer classification
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
- Keywords
- differential abundance testing, single-cell RNA sequencing, k-nearest neighbor graph, Milos, false discovery rate control, batch effects, thymus aging, cirrhotic liver
- MeSH terms
- Animals, Software, Cluster Analysis, Sequence Analysis, RNA, Gene Expression Profiling, Mice, Single-Cell Analysis
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-05.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-05.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
