Gene Set Enrichment Analysis Tools: Choosing the Right One

Gene set enrichment analysis (GSEA) is a computational method that determines whether predefined sets of genes show statistically significant, concordant differences between two biological states. Researchers use GSEA to interpret omics data by identifying biological pathways, processes, and functions that are overrepresented in a gene list or expression dataset. This article compares widely used GSEA tools including clusterProfiler, the original GSEA software, and Enrichr, with attention to their features, input formats, and statistical methods. The goal is to help students, researchers, analysts, and life-science professionals select an appropriate tool for their specific data type, research question, and computational skill level.

Understanding Gene Set Enrichment Analysis and Its Role in Omics Research

Gene set enrichment analysis addresses a fundamental challenge in omics research: after generating a list of differentially expressed genes or a ranked gene list, how does a researcher determine which biological functions are relevant? Individual gene lists often contain hundreds or thousands of entries, and manual interpretation is impractical. GSEA tools systematically compare the observed data against curated collections of gene sets that represent known biological pathways, cellular components, molecular functions, and regulatory programs.

The method has become standard practice in transcriptomics studies. A study on cervical cancer progression screened three gene expression profiles from the Gene Expression Omnibus database, identified 476 differentially expressed genes, and performed Gene Ontology and KEGG pathway enrichment analyses followed by GSEA to reveal key pathways and core genes [5]. Similarly, research on mismatch repair-deficient colorectal cancer in rhesus macaques used systems biology tools to perform GSEA for pathway discovery alongside consensus molecular subtyping and somatic mutation profiling [10]. These examples illustrate how GSEA functions as an interpretive layer between raw omics data and biological insight.

GSEA differs from simpler over-representation analysis approaches. Over-representation analysis tests whether genes from a predefined set appear more frequently in a list of differentially expressed genes than expected by chance. Functional class scoring methods, which include GSEA, consider the distribution of expression changes across all genes in a set instead of relying solely on a threshold for differential expression [13]. This distinction matters for study design because the two approaches answer different questions and have different statistical properties.

Core Principles of Gene Set Enrichment Analysis

Gene Set Collections and Annotation Databases

Gene set collections are the foundation of any enrichment analysis. The most frequently used classification mechanism for gene set definition is Gene Ontology annotation [6]. Gene Ontology organizes genes into three categories: biological processes, cellular components, and molecular functions. The Kyoto Encyclopedia of Genes and Genomes provides pathway annotations that describe molecular interaction networks. Reactome offers curated pathway information with a focus on human biology [13].

The choice of gene set collection influences results in ways that researchers should understand. A study of RhoA function applied GSEA with three collections: hallmark, KEGG pathway, and Gene Ontology Biological Process [11]. The researchers found that gene sets with similar names across collections contained extremely heterogeneous gene elements, meaning the type of collection chosen and the analytical context influence interpretation. Using multiple collections restricted the leakage of information extracted and enabled robust and consistent function identification [11].

Specialized databases expand the range of available gene sets beyond the common options. PlantGSAD collected and defined 236,007 gene sets across nine main categories for 44 plant species, covering trait ontology, co-expression modules, chromatin states, and liquid-liquid phase separation [9]. PANGEA provides gene sets for pathway annotation, protein complex data, expression and disease annotation from the Alliance of Genome Resources, with particular strength for Drosophila and other major model organisms [6]. Researchers working with non-model organisms should verify that their chosen tool supports the relevant annotation database.

Statistical Methods in GSEA Tools

GSEA tools implement different statistical approaches, and the choice affects both results and interpretation. The original GSEA method ranks all genes based on their differential expression between two conditions, then calculates an enrichment score that reflects whether members of a gene set cluster at the top or bottom of the ranked list. The significance of this score is assessed through permutation testing.

Functional class scoring methods share a common structure: compute a gene-level statistic, aggregate gene-level statistics into a pathway-level statistic, and assess significance [13]. The specific implementations vary. Weighted Kolmogorov-Smirnov testing offers an alternative statistical framework for GSEA that modifies how the enrichment score is calculated [23]. Single-sample methods such as single-sample GSEA and gene set variation analysis compute enrichment scores for individual samples instead of comparing groups, which enables different analytical applications.

A comparison of gene set scoring methods for tuberculosis gene signatures evaluated ssGSEA, GSVA, PLAGE, Singscore, and Zscore as alternative approaches to profile gene signature performance across 24 transcriptomic datasets [12]. The study found that for many signatures, the prediction performance of gene set scoring methods needed careful evaluation against the original trained models. This finding underscores that the statistical method embedded in a GSEA tool is not a neutral choice but affects the biological conclusions drawn.

Over-Representation Analysis versus Functional Class Scoring

Researchers should distinguish between over-representation analysis and functional class scoring because the tools optimized for each approach differ. Over-representation analysis requires a list of differentially expressed genes defined by a threshold, then tests whether gene sets are enriched in that list. Functional class scoring uses the full ranked gene list or expression matrix without requiring an arbitrary threshold [13].

The practical implications are significant. Over-representation analysis discards information about genes that do not meet the differential expression threshold, which can reduce statistical power. Functional class scoring retains this information and can detect coordinated changes in gene sets where individual genes show modest effects. A tutorial for bench scientists describes detailed procedures for both methods using three knowledge databases: Gene Ontology, KEGG, and Reactome [13]. The tutorial emphasizes that R provides a robust platform for enrichment analysis due to accessible packages and visualization tools.

At a Glance: Comparison of Major GSEA Tools

Practical Workflow for Selecting and Using a GSEA Tool

Step 1: Define the Research Question and Data Type

The first decision is whether the analysis requires over-representation analysis or functional class scoring. If the study has a clear list of differentially expressed genes and the question is which pathways are overrepresented, over-representation analysis tools such as Enrichr or clusterProfiler are appropriate. If the study has a full expression matrix and the question is which gene sets show coordinated changes across conditions, functional class scoring with the original GSEA software or clusterProfiler is more suitable.

The data type also matters. RNA-seq data typically requires gene-level summarization and normalization before GSEA. Microarray data may need platform-specific preprocessing. Single-cell RNA-seq data presents additional challenges because the sparsity and dropout characteristics require specialized handling. Researchers should confirm that their chosen tool accepts the data format they have.

Step 2: Select Gene Set Collections

The gene set collection should match the biological question. Hallmark gene sets provide coherent representations of well-defined biological states and are useful for broad hypothesis generation. KEGG pathways offer mechanistic detail for metabolic and signaling pathways. Gene Ontology terms provide functional annotation across three dimensions. The RhoA study demonstrated that using multiple collections improves robustness but also that gene sets with similar names can contain heterogeneous gene elements [11].

For species with limited annotation, specialized databases may be necessary. PlantGSAD provides gene sets for 44 plant species across categories including trait ontology and co-expression modules [9]. PANGEA allows GO analysis on different sets of annotations, for example excluding high-throughput studies, which can reduce bias in model organism research [6]. Researchers should document which collections and versions were used to ensure reproducibility.

Step 3: Prepare Input Data and Run the Analysis

Input preparation varies by tool. Enrichr accepts a simple gene list pasted into a web form. clusterProfiler accepts gene lists, ranked lists, or expression matrices within an R session. The original GSEA software requires an expression dataset file and a phenotype label file in specific formats.

The RumBall RNA-seq analysis platform provides a protocol that illustrates the full workflow: starting with FASTQ files, performing read mapping, sample normalization, statistical modeling, and gene ontology enrichment, then interpreting results with plots and tables [17]. This protocol demonstrates that GSEA is typically one step in a larger analysis pipeline, and the quality of upstream processing affects downstream enrichment results.

Step 4: Interpret Results with Appropriate Caution

GSEA results require interpretation within the context of the study design and the limitations of the method. A comparison of pairwise GSEA and single-sample approaches found that pairwise GSEA can overgeneralise biological enrichment, meaning that statistically significant signatures may not correspond to biological distinction between sample groups [7]. The dualGSEA tool was developed to address this issue by providing multiple statistics and visuals to aid interpretation [7].

The study of RhoA functions highlighted another interpretation challenge: there is a limit to deriving functions with high accuracy and reliability simply from gene set names [11]. Researchers should examine the leading-edge genes within enriched sets, consider the direction of enrichment, and validate findings through complementary analyses such as heatmap visualization, pathway diagramming, and direct acyclic graphing [11].

Detailed Tool Profiles and Tradeoffs

clusterProfiler

clusterProfiler is an R package that provides comprehensive functionality for gene set enrichment analysis. It supports both over-representation analysis and functional class scoring, and it can directly access the latest KEGG knowledge database [13]. The package integrates with the broader R ecosystem for statistical analysis and visualization.

The main advantage of clusterProfiler is its flexibility and integration. Researchers can move from differential expression analysis to enrichment analysis to visualization within a single R session. The package supports multiple gene set databases including Gene Ontology, KEGG, and Reactome. Visualization options include dot plots, term-gene network plots, enrichment map plots, ridge plots, and GSEA plots [13].

The main limitation is the requirement for R programming skills. Bench scientists without R experience may find the learning curve steep. However, annotated tutorials with detailed R scripts are available that keep bench scientists in mind and provide supportive descriptions of each task [13]. The CircRNAFlow pipeline incorporates clusterProfiler for functional characterization of identified circular RNAs, demonstrating its utility in specialized analysis workflows [15].

GSEA Software from the Broad Institute

The original GSEA software implements the weighted enrichment score method with permutation testing. It requires an expression dataset and phenotype labels defining the comparison groups. The software provides enrichment plots that display the enrichment score across the ranked gene list, with bars marking the leading-edge genes that contribute most to the enrichment signal [20][21].

The GSEA software is well suited for studies comparing two phenotypes with full expression matrices. The output includes enrichment scores, normalized enrichment scores, nominal p-values, and false discovery rate q-values. The enrichment plots provide a visual representation of the distribution of gene set members within the ranked list [20][21].

The main limitation is the input format requirement. The software expects specific file formats for expression data and phenotype labels, which may require preprocessing. The software is desktop-based and may not integrate as smoothly with other analysis tools as R-based alternatives.

Enrichr

Enrichr is a web-based tool that performs over-representation analysis on gene lists. It provides access to a large number of gene set libraries and presents results in an accessible interface. The tool is particularly useful for quick analyses and hypothesis generation.

A study exploring the neurobiology of Alzheimer's disease used the Enrichr platform to scrutinize COVID-19 datasets for significant overlapping enriched biological networks [19]. The study performed GSEA on an Alzheimer's disease gene signature and used Enrichr to identify enriched biological networks. This example illustrates how Enrichr can be used for exploratory analysis of gene signatures.

The main advantage of Enrichr is accessibility. No programming is required, and results are presented in a web browser with interactive features. The main limitation is that it performs over-representation analysis only, not functional class scoring. Researchers who need to analyze full expression matrices should use a different tool.

PANGEA

PANGEA is a gene set enrichment tool developed for Drosophila and common research organisms. It allows GO analysis on different sets of GO annotations, for example excluding high-throughput studies, which provides flexibility in controlling annotation quality [6]. Beyond GO, PANGEA provides gene sets for pathway annotation and protein complex data from various resources, as well as expression and disease annotation from the Alliance of Genome Resources [6].

The tool offers enhanced visualizations including a network view of gene set to gene relationships. It also allows comparison of multiple input gene lists with accompanying visualization tools for quick and easy comparison [6]. This feature is valuable for studies that need to compare enrichment across multiple conditions or experiments.

The main limitation is the focus on specific model organisms. Researchers working on species outside the supported set should verify that appropriate annotations are available.

dualGSEA

dualGSEA is an R function developed to address the problem of overgeneralization in pairwise GSEA. The tool provides users with multiple statistics and visuals to aid interpretation of results, removing the possibility of inadvertently interpreting statistical findings as equating to biological distinction between samples within groups of interest [7].

The development of dualGSEA was motivated by a comparison showing that pairwise GSEA and fGSEA provide similar results when deployed using a range of gene pre-ranking methods, but pairwise GSEA can overgeneralise biological enrichment [7]. When the most statistically significant signatures were assessed using single-sample approaches, there was a complete absence of biological distinction between these groups [7].

The main advantage of dualGSEA is its emphasis on robust interpretation. It provides a more robust basis for discovery research by allowing users to compare statistical significance alongside biological distinctions in their data [7]. The main limitation is that it is a newer tool with less established usage in the literature.

Observations and Measurements in GSEA Practice

Enrichment Scores and Leading-Edge Genes

The enrichment score is the primary output of functional class scoring methods. It reflects the degree to which a gene set is overrepresented at the top or bottom of a ranked gene list. The normalized enrichment score accounts for gene set size and enables comparison across gene sets. The false discovery rate q-value provides a measure of statistical significance adjusted for multiple testing.

Leading-edge genes are the members of a gene set that contribute most to the enrichment signal. Examining leading-edge genes can reveal the specific genes driving the enrichment and provide biological insight beyond the gene set name. Enrichment plots display the enrichment score across the ranked list with bars marking the leading-edge genes [20][21]. Fold-change heat maps for leading-edge genes provide additional context for interpreting the direction and magnitude of expression changes [20][21].

Reproducibility Considerations

Reproducibility in GSEA requires documentation of the tool version, gene set collection version, parameters, and input data processing steps. The comparison of tuberculosis gene signatures highlighted the challenge of reproducibility when original training data are unavailable and software tools or source code implementing models are inadequate [12]. The study provided an open-access software implementation of the original models for all 19 signatures for future use [12].

Researchers should record the exact parameters used for each analysis, including the ranking metric, permutation number, and gene set size filters. Version control for both the analysis code and the gene set collections is essential. The FAIR Guiding Principles provide a framework for making data findable, accessible, interoperable, and reusable [4]. Applying these principles to GSEA workflows supports reproducibility and enables others to verify results.

Quality Checks for GSEA Results

Several quality checks should be applied to GSEA results before drawing conclusions. First, verify that the input data passed quality control at the expression level. Second, check that the number of genes in each enriched set is reasonable, as very small or very large gene sets may produce unreliable results. Third, examine the direction of enrichment to confirm that it is biologically plausible. Fourth, compare results across multiple gene set collections to identify robust findings.

The RhoA study demonstrated that the comparison of multiple gene set collections showed that although the gene sets had similar names, the gene elements were extremely heterogeneous [11]. This finding supports the practice of using multiple collections and examining the gene-level composition of enriched sets instead of relying solely on set names.

Common Failure Patterns in GSEA

Input Format Errors

A frequent cause of failed GSEA runs is incorrect input formatting. Different tools expect different formats for gene identifiers, expression values, and phenotype labels. Gene identifier formats vary across platforms and species, and mismatches between the input data and the gene set collection can result in few or no genes being mapped. Researchers should verify that gene identifiers match the format used in the gene set collection and convert identifiers if necessary.

Gene Set Coverage Problems

Gene set collections vary in their coverage of different species and gene types. A gene set collection developed primarily for human genes may have poor coverage for a non-model organism. The PlantGSAD database was developed to address this gap for plant species, providing gene sets that cover all genes in Arabidopsis and Oryza sativa [9]. Researchers working with less common species should check the coverage of their chosen gene set collection before running the analysis.

Misinterpretation of Statistical Significance

A common failure pattern is interpreting statistical significance as biological significance. The dualGSEA study found that pairwise GSEA can overgeneralise biological enrichment, with statistically significant signatures showing no biological distinction between groups when assessed using single-sample approaches [7]. Researchers should validate enrichment findings with complementary analyses and consider whether the observed enrichment has biological meaning in the context of the study.

Over-Reliance on Gene Set Names

Gene set names can be misleading. The RhoA study found that gene sets with similar names across collections contained extremely heterogeneous gene elements, and there was a limit to deriving functions with high accuracy and reliability simply from gene set names [11]. Researchers should examine the genes within enriched sets and consider the biological context instead of accepting set names at face value.

Records and Documentation for GSEA Analyses

Maintaining detailed records of GSEA analyses supports reproducibility and enables troubleshooting. The following records should be maintained for each analysis:

• Tool name and version
• Gene set collection names and versions
• Input data file names and processing steps
• Parameters used, including ranking metric and permutation number
• Number of genes mapped to the gene set collection
• Enrichment scores, normalized enrichment scores, and q-values for all gene sets
• Leading-edge genes for enriched sets
• Date of analysis and analyst name

The RumBall protocol provides an example of structured documentation for RNA-seq analysis, describing steps for software setup, obtaining data, read mapping, sample normalization, statistical modeling, and gene ontology enrichment [17]. Similar documentation practices should be applied to GSEA analyses.

Limitations and Interpretation Boundaries

Statistical Limitations

GSEA methods have inherent statistical limitations. Permutation-based significance testing can be computationally intensive and may produce unstable results with small sample sizes. The choice of ranking metric and gene set size filters can affect results. Single-sample methods have different statistical properties than group comparison methods, and the choice between them should be guided by the research question.

The comparison of gene set scoring methods for tuberculosis signatures found that for many signatures, the prediction performance of gene set scoring methods needed careful evaluation against the original trained models [12]. This finding suggests that gene set scoring methods may not always serve as a reasonable proxy for the performance of the original trained model, and researchers should be cautious when using enrichment scores for prediction purposes.

Biological Interpretation Limits

GSEA identifies statistical enrichment, not causal relationships. An enriched pathway may be a downstream consequence instead of a driver of the observed phenotype. The cervical cancer study identified key KEGG pathways and related core genes through GSEA, but the findings required validation through additional analyses including protein-protein interaction networks and survival analysis [5]. Similarly, the rhesus macaque colorectal cancer study used GSEA for pathway discovery alongside consensus molecular subtyping and somatic mutation profiling [10].

The WheatCENet platform integrates GSEA with co-expression network analysis, functional annotations, and motif identification to provide a more complete picture of gene function [8]. This integrative approach addresses the limitation that GSEA alone provides limited information about regulatory mechanisms.

Species and Annotation Limitations

GSEA results depend on the quality and completeness of gene set annotations. Species with extensive annotation resources, such as human, mouse, and Drosophila, support more reliable analyses than species with limited annotation. The PANGEA tool was developed to allow more flexible and configurable approaches to data analysis using a variety of classification sets for Drosophila and other major model organisms [6]. Researchers working with less annotated species should expect reduced coverage and interpret results with caution.

Safety and Regulatory Context for GSEA in Research

Data Sharing and Genomic Data Policies

GSEA analyses often use genomic data that may be subject to data sharing policies. The NIH Genomic Data Sharing Policy provides expectations for the sharing of genomic data generated through NIH-funded research [3]. Researchers should be aware of the requirements for data deposition, access, and sharing that apply to their studies.

The FAIR Guiding Principles provide a framework for making data findable, accessible, interoperable, and reusable [4]. Applying these principles to GSEA workflows supports reproducibility and enables others to verify results. Researchers should consider how their analysis inputs and outputs can be shared in accordance with these principles and applicable policies.

Data Resources and Quality

The quality of GSEA results depends on the quality of the underlying data. NCBI provides data resources that support genomic research, including databases for sequences, expression data, and publications [2]. EMBL-EBI offers training and data resources for bioinformatics analysis [1]. Researchers should use reputable data sources and verify the quality of data before analysis.

Professional Escalation Criteria

Researchers should seek expert assistance when encountering specific challenges in GSEA analysis. Escalation is appropriate when:

• The proportion of genes mapped to the gene set collection is very low, suggesting identifier or annotation problems
• Results are highly sensitive to small changes in parameters, suggesting instability
• Enrichment results conflict across gene set collections without an obvious explanation
• The analysis involves a non-model organism with limited annotation resources
• The study requires regulatory approval or has clinical implications

Bioinformatics support teams, collaborators with statistical expertise, and tool developers can provide guidance for these situations.

Frequently Asked Questions

What is the difference between gene set enrichment analysis and pathway analysis?

Gene set enrichment analysis is a specific computational method that tests whether predefined gene sets show coordinated changes in expression data. Pathway analysis is a broader term that encompasses multiple approaches for interpreting omics data in the context of biological pathways, including over-representation analysis, functional class scoring, and pathway topology-based methods. GSEA is one type of pathway analysis that uses the full ranked gene list instead of a threshold-based gene list [13].

How do I choose between over-representation analysis and functional class scoring?

Over-representation analysis is appropriate when you have a defined list of differentially expressed genes and want to test which gene sets are overrepresented in that list. Functional class scoring is appropriate when you have a full expression matrix or ranked gene list and want to detect coordinated changes across all genes in a set without requiring an arbitrary threshold [13]. The choice depends on whether the research question focuses on a specific gene list or on genome-wide expression patterns.

What input formats do the major GSEA tools accept?

Enrichr accepts a simple gene list pasted into a web form. clusterProfiler accepts gene lists, ranked lists, or expression matrices within an R session. The original GSEA software requires an expression dataset file and a phenotype label file in specific formats. PANGEA accepts gene lists with configurable annotation options [6]. Researchers should verify that their data can be formatted appropriately for the chosen tool.

How do I choose gene set collections for my analysis?

The choice of gene set collections should match the biological question and the species under study. Hallmark gene sets provide broad coverage of well-defined biological states. KEGG pathways provide mechanistic detail for metabolic and signaling pathways. Gene Ontology terms provide functional annotation across biological processes, cellular components, and molecular functions [13]. Using multiple collections can improve robustness, but researchers should be aware that gene sets with similar names can contain heterogeneous gene elements [11].

What is the difference between pairwise GSEA and single-sample GSEA?

Pairwise GSEA compares gene expression between two groups and identifies gene sets that are enriched in one group relative to the other. Single-sample GSEA computes an enrichment score for each individual sample, which enables analysis of continuous phenotypes, heterogeneous samples, and single-cell data. The dualGSEA study found that pairwise GSEA can overgeneralise biological enrichment, and single-sample approaches can reveal a lack of biological distinction between groups that pairwise analysis suggests are different [7].

How do I interpret enrichment scores and q-values?

The enrichment score reflects the degree to which a gene set is overrepresented at the top or bottom of a ranked gene list. The normalized enrichment score accounts for gene set size and enables comparison across gene sets. The false discovery rate q-value provides a measure of statistical significance adjusted for multiple testing. Researchers should examine leading-edge genes within enriched sets and consider the direction of enrichment instead of relying solely on q-values [20][21].

Can I use GSEA for non-model organisms?

GSEA can be used for non-model organisms if appropriate gene set annotations are available. Specialized databases such as PlantGSAD provide gene sets for 44 plant species [9], and PANGEA provides annotations for Drosophila and other major model organisms [6]. For species with limited annotation resources, researchers should expect reduced coverage and interpret results with caution.

How do I ensure my GSEA results are reproducible?

Reproducibility requires documentation of the tool version, gene set collection version, parameters, and input data processing steps. The FAIR Guiding Principles provide a framework for making data findable, accessible, interoperable, and reusable [4]. Researchers should maintain detailed records of each analysis and consider sharing analysis code and processed data to enable verification by others.

Related Bioinformatics Guides

• Gene Ontology (GO) and Enrichment Analysis
• Epigenetics and Computational DNA Methylation Analysis: Mechanisms, Methods, and Veterinary Applications
• In Silico Analysis of Spike Protein Mutations and Their Impact on Host Receptor Affinity: A Computational Virology Approach
• MicroRNA Target Prediction Tools
• The KEGG Database and Pathway Analysis

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• Bioinformatics analysis of differentially expressed genes and pathways in the development of cervical cancer.. BMC cancer, 2021.
• PANGEA: a new gene set enrichment tool for Drosophila and common research organisms.. Nucleic acids research, 2023.
• Dual gene set enrichment analysis (dualGSEA), an R function that enables more robust biological discovery and pre-clinical model alignment from transcriptomics data.. Scientific reports, 2024.
• WheatCENet: A Database for Comparative Co-expression Networks Analysis of Allohexaploid Wheat and Its Progenitors.. Genomics, proteomics & bioinformatics, 2023.
• PlantGSAD: a comprehensive gene set annotation database for plant species.. Nucleic acids research, 2022.
• Comparative molecular genomic analyses of a spontaneous rhesus macaque model of mismatch repair-deficient colorectal cancer.. PLoS genetics, 2022.
• Discovery of Cellular RhoA Functions by the Integrated Application of Gene Set Enrichment Analysis.. Biomolecules & therapeutics, 2022.
• Comparison of gene set scoring methods for reproducible evaluation of multiple tuberculosis gene signatures.. bioRxiv : the preprint server for biology, 2023.
• Biological Functional Class Enrichment Analysis with R, an Annotated Tutorial for Bench Scientists.. 2026.
• A comprehensive toolkit for analyzing cell-free DNA genomic sequencing data in liquid biopsy.. 2026.
• Circular RNA Identification and Characterization with CircRNAFlow: A Bioinformatics Approach.. 2025.
• A CSF disease-associated macrophage signature defines progressive multiple sclerosis.. 2026.
• Protocol for identifying differentially expressed genes using the RumBall RNA-seq analysis platform.. 2024.
• A core stemness-associated module reveals PLK1, NUF2, KIF23, CDCA8, TOP2A, CENPF, AURKA, and ASPM as key genes in rectal cancer. European Journal of Medical Research, 2025.
• FYN, SARS-CoV-2, and IFITM3 in the neurobiology of alzheimer's disease. 2021.
• GSEA of liver transcripts in turkeys fed 2 or 1 μg Se/g vs. 0.4 μg Se/g.. 2020.
• GSEA of liver transcripts in high-Se turkeys fed 5 μg Se/g vs. 0.4 μg Se/g.. 2020.
• BubbleGUM: Automatic extraction of phenotype molecular signatures and comprehensive visualization of multiple Gene Set Enrichment Analyses. BMC Genomics, 2015.
• Weighted kolmogorov smirnov testing: An alternative for gene set enrichment analysis. Statistical Applications in Genetics and Molecular Biology, 2015.
• Transcriptomic data from two primary cell models stimulating human monocytes suggest inhibition of oxidative phosphorylation and mitochondrial function by N. meningitidis which is partially up-regulated by IL-10. BMC Immunology, 2017.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.