Glossary

Cell Cycle Scoring

A computational method that assigns each cell a numerical score representing its progression through the cell cycle phases, used to regress out cell-cycle-driven heterogeneity during preprocessing.

Get in touch Learn more

Developer demonstrating multi-agent tool use, agent tool selection interface on laptop, casual tech demo moment.

COMPUTATIONAL PREPROCESSING

What is Cell Cycle Scoring?

A computational method that assigns each cell a numerical score representing its progression through the cell cycle phases, used to regress out cell-cycle-driven heterogeneity during preprocessing.

Cell cycle scoring is a computational preprocessing method that assigns each single cell a numerical score quantifying its progression through the S and G2/M phases of the cell cycle. By correlating a cell's gene expression profile with curated sets of phase-specific marker genes, the algorithm generates a continuous metric that distinguishes actively cycling cells from quiescent ones, enabling researchers to identify and mitigate this dominant source of technical variation.

During analysis, these scores are typically regressed out alongside other technical covariates to prevent cell cycle state from dominating unsupervised clustering and pseudotime trajectory inference. This ensures that downstream dimensionality reduction and differential expression analyses reflect genuine biological heterogeneity—such as cell type identity or disease state—rather than the transient transcriptional programs associated with DNA replication and mitotic division.

COMPUTATIONAL METHOD

Key Features of Cell Cycle Scoring

Cell cycle scoring assigns each cell a numerical value representing its progression through the cell cycle, enabling the regression of cell-cycle-driven heterogeneity during single-cell preprocessing.

Core Biological Basis

Cell cycle scoring relies on the cyclical expression of phase-specific marker genes. Genes like MKI67 peak during G2/M phase, while PCNA and RRM2 are elevated during S phase. By measuring the average expression of curated gene sets, algorithms assign a G2/M score and an S score, positioning each cell along a continuum from G1 through G2/M. This captures the transcriptional signature of proliferation without requiring direct measurement of DNA content.

Seurat CellCycleScoring Implementation

The Seurat R package provides a built-in CellCycleScoring() function that uses a predefined list of canonical marker genes derived from Tirosh et al. (2016). The algorithm:

Calculates the average expression of S-phase and G2/M-phase gene sets for each cell
Subtracts the mean expression of a randomly selected background gene set to control for library complexity
Assigns a phase classification (G1, S, G2/M) based on the maximum score
Stores the S.Score and G2M.Score as metadata columns for downstream regression

EXPLORE

Regression During Preprocessing

After scoring, cell cycle heterogeneity is regressed out during data scaling to prevent it from dominating dimensionality reduction and clustering. Using Seurat's ScaleData() function with vars.to.regress = c('S.Score', 'G2M.Score'), the expression matrix is linearly adjusted to remove variance attributable to cell cycle phase. Alternatively, the difference between S and G2/M scores can be regressed as a single variable. This ensures that downstream PCA and UMAP embeddings reflect biological identity rather than proliferative status.

Alternative: scran Cyclone Method

The scran Bioconductor package implements an alternative approach called cyclone() that uses pairs of marker genes where one gene is expected to be expressed earlier in the cycle than the other. By comparing relative expression within each pair, the algorithm computes probabilities for each phase without relying on absolute expression thresholds. This method is particularly robust to differences in sequencing depth and is commonly used in the SingleCellExperiment ecosystem alongside scater and scuttle.

EXPLORE

When to Apply and When to Avoid

Cell cycle regression is essential when analyzing quiescent versus proliferating cell populations where cycling genes would otherwise define spurious clusters. However, it should be avoided or carefully evaluated when:

The biological process of interest is directly linked to proliferation, such as in tumor progression studies
Analyzing developing tissues where cell cycle state is a meaningful biological signal
Working with terminally differentiated cells like neurons that do not cycle In these cases, the cell cycle scores should be retained as a biological variable rather than regressed out.

Quality Control Interpretation

Cell cycle scores serve as a quality control metric beyond their use in regression. A healthy single-cell dataset typically shows a continuous distribution of cells across all phases, with the majority in G1. Unexpected patterns—such as a dataset dominated by S-phase cells—may indicate dissociation-induced stress or tissue-specific proliferation artifacts. Visualizing the distribution of S.Score and G2M.Score as a scatter plot colored by phase assignment provides immediate insight into data quality before proceeding with downstream analysis.

CELL CYCLE SCORING

Frequently Asked Questions

Clear, technical answers to common questions about the computational methods used to quantify and regress out cell cycle-driven heterogeneity in single-cell transcriptomic data.

Cell cycle scoring is a computational method that assigns each cell a numerical score representing its progression through the cell cycle phases (G1, S, G2/M) based on the expression of a predefined set of marker genes. The process works by calculating the average relative expression of two gene sets: one specific to the S phase (DNA synthesis) and one specific to the G2/M phase (mitosis). The difference between these scores classifies a cell's position. This is a critical preprocessing step in single-cell RNA-seq analysis because cell cycle-driven transcriptional heterogeneity can dominate the variance, masking genuine biological signals like differentiation or disease state. The most widely used implementation is the CellCycleScoring() function in the Seurat R package, which uses a curated list of 43 S-phase and 54 G2/M-phase genes derived from a study by Tirosh et al. (2016).

Enabling Efficiency, Speed & Accuracy

Intelligent Analysis, Decision & Execution

We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.

Talk to Us

Search across company data

Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.

Useful when people spend too long searching or get different answers from different systems.

Enterprise searchRAGPermissions

Automate internal workflows

Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.

Useful when repetitive work moves across multiple tools and teams.

AI agentsWorkflow automationGovernance

Add AI to products and internal tools

Build assistants, guided actions, or decision support into the software your team or customers already use.

Useful when AI needs to be part of the product, not a separate tool.

AI integrationDecision supportModel routing

COMPARATIVE ANALYSIS

Cell Cycle Scoring vs. Related Methods

A comparison of computational approaches for identifying and mitigating cell-cycle-driven heterogeneity in single-cell RNA-seq data.

Feature	Cell Cycle Scoring	Regression (e.g., Seurat)	scVI Latent Correction
Primary Mechanism	Assigns phase scores based on marker gene expression	Regresses out cell cycle gene scores during data scaling	Models cell cycle as a nuisance latent variable in a deep generative framework
Input Required	List of S and G2/M phase marker genes	Cell cycle phase scores (S.Score, G2M.Score)	Raw count matrix with batch annotations
Preserves Biological Heterogeneity
Handles Zero-Inflation
Nonlinear Effect Modeling
Computational Cost	Low	Low	High (GPU recommended)
Interpretability	High (direct gene-phase linkage)	Medium (variance removed is quantifiable)	Low (latent space is abstract)
Risk of Over-Correction	Low (only classifies, does not subtract signal)	High (may remove biological signal correlated with cell cycle)	Medium (depends on model specification)

SINGLE-CELL WORKFLOW

Related Terms

Cell cycle scoring is a critical preprocessing step. Explore the interconnected methods that prepare, correct, and interpret single-cell data after this heterogeneity is quantified.

Count Matrix Normalization

A preprocessing step that adjusts raw gene expression counts to account for differences in sequencing depth and capture efficiency between cells. Normalization is essential before cell cycle scoring to ensure that technical artifacts do not confound the biological signal of proliferation. Common methods include library-size scaling and regularized negative binomial regression.

Data Integration

The computational alignment of multiple single-cell datasets into a shared latent space. After cell cycle scoring, the calculated phase scores are often used as variables to regress out during integration. This prevents cell-cycle-driven variation from dominating the alignment and masking true biological differences between conditions, donors, or technologies.

Highly Variable Gene Selection

A feature selection method that identifies genes with the highest cell-to-cell variation. This step often explicitly excludes cell cycle genes to prevent them from dominating the downstream dimensionality reduction. By focusing on genes that vary due to biological identity rather than proliferation, the resulting clusters reflect cell types instead of mitotic phase.

Pseudotime Trajectory Inference

A computational ordering of cells along a continuous developmental path. Cell cycle scoring is crucial here to distinguish differentiation trajectories from cyclic proliferative signals. If cell cycle effects are not regressed, trajectory inference algorithms may incorrectly place cycling progenitor cells along a false temporal axis that merely reflects the G1, S, or G2/M phases.

scVI

Single-cell Variational Inference, a deep generative model that learns a probabilistic latent representation of gene expression. scVI can natively account for cell cycle scores as a nuisance covariate during training. By modeling the expression distribution conditioned on cycle phase, it effectively removes the proliferative signal from the latent space without a separate linear regression step.

SCENIC

Single-Cell rEgulatory Network Inference and Clustering. This method identifies active transcription factors and their target regulons. Cell cycle scoring is often cross-referenced with SCENIC output to validate that the inferred regulators of proliferation, such as E2F family members, are correctly assigned to cells with high S-phase or G2/M scores, confirming the biological accuracy of the computational pipeline.

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.

Limited slotsGet a Free AI Consultation

How We Work

Custom AI workflows for your Business

One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.