CoChat

CellDrift: inferring perturbation responses in temporally sampled single-cell data

Open access

Published version of a preprint

The DOI matches its Crossref record; 53 of the 54 references it cites were found in a registry or an index.

Abstract

Cells and tissues respond to perturbations in multiple ways that can be sensitively reflected in the alterations of gene expression. Current approaches to finding and quantifying the effects of perturbations on cell-level responses over time disregard the temporal consistency of identifiable gene programs. To leverage the occurrence of these patterns for perturbation analyses, we developed CellDrift (https://github.com/KANG-BIOINFO/CellDrift), a generalized linear model-based functional data analysis method that is capable of identifying covarying temporal patterns of various cell types in response to perturbations. As compared to several other approaches, CellDrift demonstrated superior performance in the identification of temporally varied perturbation patterns and the ability to impute missing time points. We applied CellDrift to multiple longitudinal datasets, including COVID-19 disease progression and gastrointestinal tract development, and demonstrated its ability to identify specific gene programs associated with sequential biological processes, trajectories and outcomes.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/bib/bbac324

Cites

53 of 53 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Workflow of CellDrift. (A) An example of a perturbational single-cell experiment with multiple time points. (B) Real scenarios of single-cell experiments with varying perturbation effects over time. (C) GLM with the interaction of cell-type-perturbation applied separately at each time point, and contrast coefficients are derived as the representation of perturbation responses. The complete model can be found in the Methods. (D) Contrast coefficients are used as input for various applications in FDA.
Figure 1 Workflow of CellDrift. (A) An example of a perturbational single-cell experiment with multiple time points. (B) Real scenarios of single-cell experiments with varying perturbation effects over time. (C) GLM with the interaction of cell-type-perturbation applied separately at each time point, and contrast coefficients are derived as the representation of perturbation responses. The complete model can be found in the Methods. (D) Contrast coefficients are used as input for various applications in FDA.CC BY-NC · PubMed Central
Figure 2 Performance of CellDrift in the identification of perturbed genes and temporal perturbation patterns. (A, B) Benchmark results of the GLM of CellDrift and other commonly used differential expression approaches in the simulated data. Simulated data with multiple batch effect sizes (A) and differential expression sizes (B) were used for benchmarking. TPR and FDR were used as metrics. The mean and SD derived from 10 replicates for each test are shown in the figure. (C) Benchmark results of temporal pattern recovery for CellDrift strategy (GLM + Fuzzy C-mean) and other approaches. Temporal pattern recovery was evaluated with ARI and measured across varying parameters, including numbers of time points, ratios of missing time points and noise levels. (D) Benchmark results of imputation performance using Pearson correlations between imputed contrast coefficients from FPCA and real simulated perturbation coefficients. The same parameters were used as (C).
Figure 2 Performance of CellDrift in the identification of perturbed genes and temporal perturbation patterns. (A, B) Benchmark results of the GLM of CellDrift and other commonly used differential expression approaches in the simulated data. Simulated data with multiple batch effect sizes (A) and differential expression sizes (B) were used for benchmarking. TPR and FDR were used as metrics. The mean and SD derived from 10 replicates for each test are shown in the figure. (C) Benchmark results of temporal pattern recovery for CellDrift strategy (GLM + Fuzzy C-mean) and other approaches. Temporal pattern recovery was evaluated with ARI and measured across varying parameters, including numbers of time points, ratios of missing time points and noise levels. (D) Benchmark results of imputation performance using Pearson correlations between imputed contrast coefficients from FPCA and real simulated perturbation coefficients. The same parameters were used as (C).CC BY-NC · PubMed Central
Figure 3 Temporal perturbation effects in COVID-19 atlas. (A) Overview of the number of cells in each cell group of the dataset, which contains multiple disease conditions, cell types and time points from days 1 to 25 since the disease onset. The size of dots represents the number of cells. HCW_MILD: healthcare workers with mild COVID-19; MILD, SEV, CRIT: mild, severe and critical COVID-19; CD4, CD8: CD4 T cell, CD8 T cell; cMono, ncMono: classical and non-classical monocyte. (B) Three distinct temporal patterns of contrast coefficients from classical monocytes of severe COVID-19 patients. The top row shows curves of genes with similar contrast coefficients in each cluster over time, and the bottom row shows the gene set enrichment analysis of genes in each temporal cluster. The black line represents the average time curve for all genes, which is the same across three plots. Gene enrichment scores are defined as }{}$-{\log}_{10}\mathrm{FDR}$-adjusted P-values of enrichment significance. (C) Five genes from cluster 17 were prioritized by the functional ANOVA test, which have significantly higher temporal curves in severe conditions than mild symptoms. Contrast coefficients for classical monocytes across disease conditions are shown on smoothed curves computed by FPCA, and time curves were aligned using dynamic time warping. (D) Validation of genes from (C) with another large-scale COVID-19 PBMC data [1]. FPCA smoothed curves for genes in three replicates of mild and severe patients are shown, which display similar temporal patterns as (C).
Figure 3 Temporal perturbation effects in COVID-19 atlas. (A) Overview of the number of cells in each cell group of the dataset, which contains multiple disease conditions, cell types and time points from days 1 to 25 since the disease onset. The size of dots represents the number of cells. HCW_MILD: healthcare workers with mild COVID-19; MILD, SEV, CRIT: mild, severe and critical COVID-19; CD4, CD8: CD4 T cell, CD8 T cell; cMono, ncMono: classical and non-classical monocyte. (B) Three distinct temporal patterns of contrast coefficients from classical monocytes of severe COVID-19 patients. The top row shows curves of genes with similar contrast coefficients in each cluster over time, and the bottom row shows the gene set enrichment analysis of genes in each temporal cluster. The black line represents the average time curve for all genes, which is the same across three plots. Gene enrichment scores are defined as }{}$-{\log}_{10}\mathrm{FDR}$-adjusted P-values of enrichment significance. (C) Five genes from cluster 17 were prioritized by the functional ANOVA test, which have significantly higher temporal curves in severe conditions than mild symptoms. Contrast coefficients for classical monocytes across disease conditions are shown on smoothed curves computed by FPCA, and time curves were aligned using dynamic time warping. (D) Validation of genes from (C) with another large-scale COVID-19 PBMC data [1]. FPCA smoothed curves for genes in three replicates of mild and severe patients are shown, which display similar temporal patterns as (C).CC BY-NC · PubMed Central
Figure 4 Differential temporal trajectories of gut development. (A) Overview of fetal gut cell atlas. The study contains single-cell sequencing data for three compartments in gut development, including duo-jejunum, ileum and colon, whose transcriptomics profiles were retrieved from weeks 6 to 11. (B) The top 2 FPCA eigenfunctions and their scores over time for functional data of comparisons of colon and duo-jejunum in mesenchymal cells. First eigenfunction }{}${\phi}_1$ presents a strong temporal pattern as is highlighted. (C) Functional principal component 1 (FPC1) scores were computed for genes in each cluster, displayed in decreasing order. High positive scores indicate genes in the cluster may have similar temporal patterns as }{}${\phi}_1$, while large negative scores indicate opposite temporal patterns as }{}${\phi}_1$. Clusters 11 and 16 are highlighted for the following analysis. (D, E) FPCA smoothed curves for genes in cluster 11 (D) and 16 (E) and their gene enrichment results. (F) The top three FPCA eigenfunctions and their scores over time for functional data of comparisons of colon and duo-jejunum in epithelium cells. The second eigenfunction }{}${\phi}_2$ presents a strong temporal pattern as is highlighted. (G) Functional principal component 2 (FPC2) scores were computed for genes in each cluster, which was ranked in a decreasing order. Clusters 4 and 9 are highlighted for the following analysis. (H, I) FPCA smoothed curves for genes in clusters 9 (H) and 4 (I) and their gene enrichment results.
Figure 4 Differential temporal trajectories of gut development. (A) Overview of fetal gut cell atlas. The study contains single-cell sequencing data for three compartments in gut development, including duo-jejunum, ileum and colon, whose transcriptomics profiles were retrieved from weeks 6 to 11. (B) The top 2 FPCA eigenfunctions and their scores over time for functional data of comparisons of colon and duo-jejunum in mesenchymal cells. First eigenfunction }{}${\phi}_1$ presents a strong temporal pattern as is highlighted. (C) Functional principal component 1 (FPC1) scores were computed for genes in each cluster, displayed in decreasing order. High positive scores indicate genes in the cluster may have similar temporal patterns as }{}${\phi}_1$, while large negative scores indicate opposite temporal patterns as }{}${\phi}_1$. Clusters 11 and 16 are highlighted for the following analysis. (D, E) FPCA smoothed curves for genes in cluster 11 (D) and 16 (E) and their gene enrichment results. (F) The top three FPCA eigenfunctions and their scores over time for functional data of comparisons of colon and duo-jejunum in epithelium cells. The second eigenfunction }{}${\phi}_2$ presents a strong temporal pattern as is highlighted. (G) Functional principal component 2 (FPC2) scores were computed for genes in each cluster, which was ranked in a decreasing order. Clusters 4 and 9 are highlighted for the following analysis. (H, I) FPCA smoothed curves for genes in clusters 9 (H) and 4 (I) and their gene enrichment results.CC BY-NC · PubMed Central
Figure 5 Computational performance of CellDrift. (A, B) Total amount of running time (A) and memory usage (B) of the CellDrift application in single-cell datasets with various numbers of cells and genes. The mean and standard error values were calculated for three runs in each condition.
Figure 5 Computational performance of CellDrift. (A, B) Total amount of running time (A) and memory usage (B) of the CellDrift application in single-cell datasets with various numbers of cells and genes. The mean and standard error values were calculated for three runs in each condition.CC BY-NC · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

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?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 53 references

Cited by

Metadata

Authors
Jin, K., Schnell, D., Li, G., Salomonis, N., Prasath, V. B. S., Szczesniak, R., & Aronow, B. J.
Affiliations
Cincinnati Children's Hospital Medical Center (US)
Corresponding author
Aronow, Bruce J
Venue
Briefings in Bioinformatics
Published
2022-08-23
Volume
23
Issue
5
ISSN
  • 1467-5463 (ISSN-L)
  • 1477-4054
OpenAlex
W4293023608
Cited by
11
References
54
Citations by year
2026: 2 · 2025: 5 · 2024: 2 · 2023: 2
FWCI
0.73
Type
journal article
Language
en
Open access
Locations
  • Briefings in Bioinformatics (publishedVersion, cc-by-nc)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
Funders
  • National Institutes of Health (HL148865)
  • Digestive Health Center (DK078392)
  • National Cooperative Reprogrammed Cell Research Program (MH104172)
  • Center of Excellence in Molecular Hematology (DK126108)
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 › Gene Regulatory Network Analysis
Keywords
single-cell trajectory analysis, temporal gene expression, functional data analysis, gene programs, gastrointestinal tract development, cell type-specific responses
MeSH terms
COVID-19, Humans, Linear Models

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.

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

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.