RNA-Seq Data Analysis in Galaxy: A User-Friendly Platform

RNA sequencing data analysis requires multiple computational tools that are often difficult to install and run for researchers without advanced programming experience. Galaxy provides a web-based platform where these tools are pre-installed, connected through a graphical interface, and accessible through any standard browser. This article explains how to perform a complete RNA-seq analysis workflow in Galaxy, from uploading raw sequencing data to visualizing differentially expressed genes, with specific attention to quality control, parameter choices, reproducibility, and interpretation limits.

At a Glance

The table below summarizes the main stages of a reference-based RNA-seq analysis in Galaxy, the tools commonly used at each stage, and the key decisions a researcher must make.

Understanding the RNA-seq Analysis Pipeline

RNA-seq analysis converts raw sequencing reads into biological insight through a series of computational steps. Each step transforms data and introduces potential sources of error. A typical reference-based workflow includes quality control, read trimming, alignment to a reference genome, quantification of gene expression, and statistical testing for differential expression. The Galaxy platform embeds these tools in a web interface and provides reproducibility through workflow management, as described in the EMBL-EBI Training materials and the NCBI Data Resources documentation.

The computational demands of RNA-seq analysis have historically created barriers for experimental biologists. Galaxy addresses this by hosting tools on high-capacity servers that are accessible for free, with an option to install the platform locally when data sensitivity or scale requires it. A 2021 protocol article in Current Protocols demonstrates how Galaxy provides access to computational biology tools in a web-based interface and connects to major public biological data repositories, allowing private data to be combined with public datasets. The article presents protocols for several analysis types including single-cell RNA-seq, showing the breadth of the platform beyond standard bulk RNA-seq.

For researchers who need to move between graphical interfaces and command-line environments, the Multi-Interface Galaxy Hands-on Training Suite (MIGHTS) offers parallel analytical methods using either buttons or code. A 2025 article in GigaScience describes how MIGHTS provides 11 tutorials with recordings, slide decks, and interactive visualizations, and emphasizes FAIR data principles to ensure validation of results. This approach helps bench biologists and clinicians who face a steep learning curve when learning to code while analyzing data.

Setting Up a Galaxy Account and Workspace

Before starting an RNA-seq analysis, create a free account on a public Galaxy server. The main Galaxy server at galaxyproject.org is suitable for most educational and research analyses. Some analyses require substantial computational resources, and public servers may have usage quotas. For large datasets or sensitive data, consider installing Galaxy locally or using an institutional server.

After creating an account, organize your workspace using histories. Each history contains the datasets and tool outputs for one analysis project. Create a new history for each RNA-seq experiment to keep results organized and reproducible. Name histories descriptively with the project name, date, and analysis version.

Galaxy stores every analysis step in the history, which provides a record of all tools, parameters, and inputs used. This feature supports reproducibility because anyone with access to the history can rerun the analysis with identical settings. The FAIR Guiding Principles published in Scientific Data describe the importance of making data findable, accessible, interoperable, and reusable, and Galaxy histories support these principles by capturing the full analysis trail.

Uploading RNA-seq Data to Galaxy

RNA-seq data typically arrives as FASTQ files containing raw sequencing reads. Each read includes a nucleotide sequence and a quality score for each base. Galaxy accepts uploads from local computers, from URLs, and directly from public repositories such as the NCBI Sequence Read Archive (SRA).

To upload from a local computer, use the upload button in the Galaxy toolbar and select the FASTQ files. For large files, compressed formats such as FASTQ.GZ reduce upload time and storage requirements. Galaxy automatically decompresses these files during analysis.

To import data directly from NCBI SRA, use the NCBI SRA accession number. The NCBI Data Resources portal provides access to the Sequence Read Archive, where published RNA-seq datasets are deposited. A tutorial published in CourseSource describes how students import sequence files directly from NCBI SRA into a Galaxy user account, eliminating the need for on-site file storage. The tutorial uses RNA-seq data from a published study of nematode infection of Arabidopsis thaliana and guides students through quality assessment, trimming, read mapping, counts table generation, and differential expression analysis.

When uploading multiple samples from the same experiment, use Galaxy collections to group them. Collections allow tools to process all samples with the same parameters in a single step, which saves time and reduces the chance of parameter inconsistencies between samples.

Quality Control with FastQC

Raw sequencing data contains artifacts that can affect downstream analysis. Adapter sequences, low-quality base calls, and PCR duplicates are common issues. Quality control identifies these problems before they propagate through the analysis.

FastQC is the standard tool for assessing sequencing read quality. Run FastQC on each FASTQ file to generate a quality report that includes per-base quality scores, GC content, sequence duplication levels, adapter content, and overrepresented sequences. The report uses a traffic light system where green indicates normal results, yellow indicates slightly unusual results, and red indicates very unusual results.

Interpret FastQC results with attention to the following modules:

Per-base sequence quality shows the distribution of quality scores across read positions. RNA-seq reads often show declining quality toward the 3 prime end. If quality drops below Q20 in a substantial fraction of reads, trimming is required.

Adapter content identifies sequences matching common adapter sequences. Adapter contamination occurs when the insert fragment is shorter than the read length. High adapter content requires adapter trimming.

Sequence duplication levels indicate the proportion of reads that are exact duplicates. High duplication can result from PCR amplification bias or from high expression of a small number of genes. For RNA-seq, some duplication is expected because highly expressed genes produce many identical reads.

Overrepresented sequences may indicate adapter contamination, rRNA contamination, or the presence of highly expressed transcripts.

After running FastQC on all samples, use MultiQC to aggregate the individual reports into a single summary. This makes it easier to compare quality across samples and identify samples that deviate from the group.

Read Trimming with Trimmomatic

Trimming removes low-quality bases and adapter sequences from reads. The Trimmomatic tool in Galaxy provides several trimming steps that can be combined in a single run.

The key parameters for Trimmomatic are:

Adapter sequences to remove. Galaxy provides preset adapter sequences for common Illumina adapters. Select the correct preset for your library preparation kit.

Quality trimming threshold. Trimmomatic uses a sliding window approach where the average quality within a window of bases must exceed a threshold. A common setting is a window of 4 bases with a minimum average quality of Q20.

Minimum read length. Reads shorter than this length after trimming are discarded. A common threshold is 36 bases, because shorter reads cannot be aligned reliably.

The choice of trimming parameters depends on the data quality and the downstream analysis. Aggressive trimming removes more bases and may discard useful data. Conservative trimming retains more data but may leave low-quality bases that affect alignment.

After trimming, rerun FastQC on the trimmed reads to confirm that quality issues have been resolved. Compare the before and after reports to document the effect of trimming.

Aligning Reads to a Reference Genome

Alignment maps each read to its location in the reference genome. For RNA-seq, the aligner must be splice-aware because reads can span exon-exon junctions. STAR and HISAT2 are the most commonly used splice-aware aligners in Galaxy.

Before alignment, select the reference genome and its annotation. Galaxy provides built-in reference genomes for many organisms. Choose the genome version that matches your experimental organism and note the version in your analysis records. The gene annotation file (GTF or GFF format) defines the gene models used for quantification.

STAR alignment parameters that affect results include:

The reference genome index. Galaxy provides pre-built indexes for common genomes. Using a pre-built index saves time and ensures consistency.

The number of mismatches allowed. The default of 10 is appropriate for most analyses. Increasing mismatches may improve alignment for divergent samples but increases false alignments.

The handling of multi-mapping reads. Reads that align to multiple locations can be counted once, counted multiple times, or discarded. The choice affects expression estimates for genes with high sequence similarity.

The output format. STAR can output BAM files for visualization and quantification. Galaxy also provides options for outputting junction files and other auxiliary files.

After alignment, check the alignment statistics. The STAR log file reports the percentage of reads that mapped uniquely, mapped to multiple loci, and did not map. A uniquely mapped percentage below 70 percent may indicate problems with the reference genome, the data quality, or the trimming parameters.

Quantifying Gene Expression with featureCounts

Quantification counts the number of reads that overlap each gene. The featureCounts tool in Galaxy is efficient and widely used for this purpose.

The key parameters for featureCounts are:

The gene annotation file. Use the same annotation file that was used for alignment to ensure consistency between gene models and read counting.

The feature type. The default is exon, which counts reads overlapping exons. Some analyses use other feature types such as gene or transcript.

The counting mode. featureCounts can count reads that overlap any exon of a gene, or reads that overlap a specific set of exons. The default mode counts reads that overlap any exon.

The strand specificity. RNA-seq libraries can be unstranded, stranded, or reverse-stranded. Select the correct setting for your library preparation kit. Using the wrong strand setting can result in counting reads on the wrong strand and losing a substantial fraction of counts.

The multi-mapping read handling. Reads that map to multiple locations can be counted or excluded. The default excludes multi-mapping reads.

The output of featureCounts is a counts table with one row per gene and one column per sample. This table is the input for differential expression analysis.

Differential Expression Analysis with DESeq2

Differential expression analysis identifies genes whose expression differs between experimental conditions. DESeq2 is a widely used tool in Galaxy that performs normalization, statistical testing, and output of results.

The input to DESeq2 is the counts table from featureCounts and a factor table that describes the experimental design. The factor table assigns each sample to a condition, such as treated or control.

DESeq2 performs several steps:

Normalization. DESeq2 estimates size factors that account for differences in sequencing depth and library composition between samples. This normalization is more robust than simple counts per million because it accounts for the fact that highly expressed genes can skew the total count.

Dispersion estimation. DESeq2 estimates the biological variability between replicates for each gene. Genes with similar expression levels are pooled to obtain more stable dispersion estimates.

Statistical testing. DESeq2 fits a negative binomial model for each gene and tests for differential expression using a Wald test or a likelihood ratio test.

The output of DESeq2 includes for each gene the log2 fold change, the mean expression, the p-value, and the adjusted p-value. The adjusted p-value accounts for multiple testing and is the primary significance measure.

The choice of significance threshold depends on the experimental context. A common threshold is an adjusted p-value below 0.05 and an absolute log2 fold change above 1. These thresholds should be justified in the analysis records.

Visualizing Results

Visualization helps interpret differential expression results and communicate findings. Galaxy provides several plotting tools that accept DESeq2 output.

A volcano plot shows the relationship between fold change and statistical significance. Each point represents a gene, with the log2 fold change on the x-axis and the negative log10 adjusted p-value on the y-axis. Genes with large fold changes and high significance appear in the upper corners of the plot.

A heatmap shows the expression levels of selected genes across samples. Heatmaps are useful for examining the expression patterns of the most significant genes or of genes in a specific pathway. Galaxy provides heatmap tools that accept the normalized counts from DESeq2.

A PCA plot shows the overall similarity between samples. Samples that cluster together have similar expression profiles. PCA plots are useful for identifying outliers and for confirming that biological replicates cluster together and that different conditions separate.

An MA plot shows the relationship between mean expression and fold change. The plot helps identify genes with large fold changes at low expression levels, which are often less reliable.

A tutorial published in Methods in Molecular Biology describes how to perform a reference-based RNA-seq analysis using Galaxy from data upload to visualization and functional enrichment analysis of differentially expressed genes. The Galaxy Training Network provides tutorials for RNA-seq analysis that include visualization steps.

Reproducibility and Workflow Management

Galaxy workflows capture the entire analysis as a reusable pipeline. A workflow defines the tools, their parameters, and the connections between them. Once a workflow is created, it can be applied to new datasets with the same parameters.

To create a workflow from an existing history, use the workflow extraction feature in Galaxy. This creates a workflow that reproduces the analysis steps in the history. The workflow can then be edited to adjust parameters or to add new steps.

Workflows support reproducibility in several ways:

They document the exact tools and parameters used in an analysis.

They allow the analysis to be rerun on new data with identical settings.

They can be shared with collaborators or published with a paper.

The FAIR Guiding Principles emphasize the importance of reproducibility in scientific research. Galaxy workflows support these principles by capturing the analysis in a form that can be reused and validated.

The Galaxy Training Network provides a large collection of tutorials that include workflows for RNA-seq analysis. A 2025 article in GigaScience describes the Galaxy single-cell and spatial omics community, which has developed more than 175 tools and 120 training resources, with more than 300,000 jobs running at the time of writing. This community effort demonstrates the scale of Galaxy usage and the value of shared training materials.

Common Failure Patterns and Troubleshooting

Several problems recur in RNA-seq analysis in Galaxy. Recognizing these patterns helps avoid wasted time and incorrect results.

Poor alignment rates often result from incorrect reference genome selection. Check that the genome version matches the organism and that the annotation file is compatible with the genome version. A mismatch between genome and annotation can cause reads to align but not be counted.

Low counts after featureCounts often result from incorrect strand settings. If the library is stranded and the wrong setting is used, reads may be counted on the wrong strand. Check the library preparation protocol and set the strand parameter accordingly.

High duplication rates in FastQC can result from PCR amplification bias or from low input RNA. If duplication is high, consider whether the library preparation protocol is appropriate for the input amount.

Samples that cluster unexpectedly in PCA plots may indicate batch effects, sample mislabeling, or contamination. Check the sample metadata and the alignment statistics for each sample.

Memory errors during STAR alignment can occur with large genomes or large numbers of reads. Galaxy public servers have memory limits. For large datasets, consider using a smaller subset of reads for initial analysis or using an institutional Galaxy server with more resources.

Limitations and Interpretation Constraints

RNA-seq analysis in Galaxy has limitations that affect interpretation of results.

Public Galaxy servers have usage quotas that limit the number of jobs and the amount of data that can be processed. Large RNA-seq experiments with many samples may exceed these quotas. Options include using an institutional server, installing Galaxy locally, or using a cloud-based deployment.

The choice of reference genome and annotation affects results. Different genome versions and annotation files can produce different counts for the same data. Results should be interpreted in the context of the specific genome version and annotation used.

Differential expression analysis identifies statistical associations, not causal relationships. A gene that is differentially expressed between conditions may be a cause, a consequence, or a bystander of the biological difference. Functional validation is required to establish biological significance.

The analysis pipeline described here is reference-based, meaning it requires a reference genome. For organisms without a reference genome, a de novo transcriptome assembly approach is needed, which is more complex and less standardized.

The Genomic Data Sharing Policy from the National Institutes of Health describes expectations for sharing genomic data generated with NIH funding. Researchers should be aware of these expectations when planning analyses and publications.

Records and Documentation

Maintain records of the analysis for reproducibility and publication. The following information should be documented:

The Galaxy server and version used.

The date of the analysis.

The reference genome version and annotation file used.

The tools and versions used at each step.

The parameters used for each tool.

The input data files and their sources.

The output files and their locations.

Galaxy histories provide most of this information automatically. Export the history as a file for inclusion in supplementary materials or for sharing with collaborators.

For published analyses, include the Galaxy workflow and the history in the supplementary materials. This allows reviewers and readers to reproduce the analysis and to verify the results.

Professional Escalation Criteria

Some situations require escalation to a bioinformatics specialist or to the Galaxy support community.

If alignment rates are consistently below 50 percent across multiple samples, the problem may be in the data generation or in the reference genome selection. Consult with the sequencing facility or with a bioinformatics specialist.

If DESeq2 produces no significant genes despite clear biological differences, the experimental design may have insufficient replication or the effect size may be small. Consult with a statistician or bioinformatics specialist about experimental design and analysis options.

If samples show unexpected clustering in PCA plots, there may be batch effects or sample mislabeling. Check the sample metadata and consider whether batch correction is needed.

If the analysis requires tools that are not available on the public Galaxy server, consult the Galaxy ToolShed or the Galaxy support community about alternative tools or server options.

The Galaxy Training Network provides a forum for asking questions and getting help from the community. The Galaxy Training Material for single-cell RNA-seq tutorial with Plant Datasets and the Training material for single-cell RNA-seq data analysis are examples of community-developed training resources that include troubleshooting guidance.

Safety and Data Handling Considerations

RNA-seq data can include human sequences that are subject to privacy and consent requirements. The Genomic Data Sharing Policy from the National Institutes of Health describes expectations for protecting research participant privacy and for responsible sharing of genomic data. Researchers working with human data should review these expectations before uploading data to public servers.

Public Galaxy servers store data on shared infrastructure. For sensitive data, use an institutional Galaxy server or install Galaxy locally. The Using Galaxy to Perform Large-Scale Interactive Data Analyses-An Update article notes that Galaxy can be installed locally, which provides control over data storage and access.

When uploading data to public servers, verify that the data can be shared under the terms of the consent and the data use agreement. Some datasets have restrictions on where they can be stored and analyzed.

Frequently Asked Questions

What is the difference between a reference-based and a de novo RNA-seq analysis?

A reference-based analysis aligns reads to an existing reference genome and counts reads that overlap annotated genes. This approach requires a high-quality reference genome and annotation for the organism. A de novo analysis assembles reads into transcripts without a reference genome, which is necessary for organisms without a reference genome. De novo assembly is more computationally intensive and produces results that are more difficult to interpret.

How do I choose between STAR and HISAT2 for alignment?

Both STAR and HISAT2 are splice-aware aligners that work well for RNA-seq data. STAR is generally faster and more sensitive, while HISAT2 uses less memory. For most analyses, either tool produces acceptable results. The choice may depend on the available computational resources and on the specific characteristics of the data. Try both tools on a subset of reads and compare the alignment statistics.

What is the minimum number of biological replicates for differential expression analysis?

Differential expression analysis requires biological replicates to estimate variability. Three replicates per condition is a common minimum, but more replicates increase statistical power and reliability. With fewer than three replicates, DESeq2 may not produce reliable dispersion estimates, and the results should be interpreted with caution.

How do I handle batch effects in RNA-seq data?

Batch effects are systematic differences between samples that are not related to the biological condition of interest. They can result from samples being processed at different times, on different sequencing runs, or by different technicians. If batch information is recorded, it can be included in the DESeq2 model as a covariate. If batch effects are unknown, PCA plots may reveal them as unexpected clustering.

Can I use Galaxy for single-cell RNA-seq analysis?

Yes, Galaxy provides tools and workflows for single-cell RNA-seq analysis. The Galaxy single-cell and spatial omics community has developed more than 175 tools and 120 training resources for this purpose. The Galaxy single-cell and spatial omics community update describes the development of this community and its resources. Single-cell analysis involves additional steps such as cell filtering, normalization, dimensionality reduction, and clustering that are not covered in the bulk RNA-seq workflow described here.

How do I share my Galaxy analysis with collaborators or reviewers?

Galaxy histories and workflows can be shared through the Galaxy interface. You can share a history or workflow with specific users or make it publicly accessible. For published analyses, export the history and workflow as files and include them in the supplementary materials. This allows others to reproduce the analysis and to verify the results.

What should I do if my alignment rate is very low?

Low alignment rates can result from several causes. Check that the reference genome matches the organism and that the genome version is correct. Check the FastQC reports for adapter contamination or low-quality bases that may have prevented alignment. Check that the trimming parameters were appropriate for the data. If the problem persists, consult with the sequencing facility or a bioinformatics specialist.

How do I choose the significance threshold for differential expression?

The significance threshold should be chosen based on the experimental context and the goals of the analysis. A common threshold is an adjusted p-value below 0.05 and an absolute log2 fold change above 1. More stringent thresholds reduce false positives but may miss true differences. The choice of threshold should be documented in the analysis records and justified in the publication.

Related Bioinformatics Guides

• Alternative Splicing Analysis from RNA-Seq Data
• Single-Cell RNA-Seq Analysis Pipelines for Veterinary Immunology
• RNA Seq Fastq Example: Structural Analysis and Computational Methodologies in Bioinformatics
• RNA-Seq Differential Expression: DESeq2, edgeR, and limma-voom Frameworks
• Single-cell RNA-seq Trajectory Inference and Cell Lineage Tracing

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.
• Using Galaxy to Perform Large-Scale Interactive Data Analyses-An Update.. Current protocols, 2021.
• Galaxy as a gateway to bioinformatics: Multi-Interface Galaxy Hands-on Training Suite (MIGHTS) for scRNA-seq.. GigaScience, 2025.
• G-OnRamp: a Galaxy-based platform for collaborative annotation of eukaryotic genomes.. Bioinformatics (Oxford, England), 2019.
• Pyicos: a versatile toolkit for the analysis of high-throughput sequencing data.. Bioinformatics (Oxford, England), 2011.
• iPepGen: a modular, immunopeptidogenomic analysis pipeline for discovery, verification, and prioritization of cancer peptide neoantigen candidates.. 2026.
• Metatranscriptomic Reanalysis of Alzheimer's Brains Identifies Low-Biomass Microbial Signals Including Enrichment of &lt,i&gt,Acinetobacter radioresistens&lt,/i&gt,.. 2026.
• CRESCENT, a comprehensive RNA-Seq expression, splicing, and coding/non-coding element network tool.. 2026.
• Galaxy single-cell & spatial omics community update: Navigating new frontiers in 2025.. 2025.
• Harnessing plasma transcriptomics for non-invasive cancer biomarker identification: a comprehensive review.. 2025.
• RAGER: A user-friendly computational platform for integrated analysis of RNA-Seq and ATAC-seq data.. 2026.
• From Prompt to Pipeline: Large Language Models for Scientific Workflow Development in Bioinformatics. 2025.
• Galaxy Training Material for single-cell RNA-seq tutorial with Plant Datasets. 2021.
• Tackling “Big Data” with Biology Undergrads: A Simple RNA-seq Data Analysis Tutorial Using Galaxy.. CourseSource, 2019.
• Training material for single-cell RNA-seq data analysis (Galaxy Training Network tutorial). 2018.
• Current best practices in single-cell RNA-seq analysis: a tutorial. Molecular Systems Biology, 2019.
• Using "Galaxy-rCASC": A Public Galaxy Instance for Single-Cell RNA-Seq Data Analysis.. Methods in molecular biology, 2023.
• RNA-Seq Data Analysis in Galaxy.. Methods in molecular biology, 2021.

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