Gene Expression
Gene expression is the process by which genetic information encoded in DNA is converted into functional products, most commonly proteins. This guide explains the core concepts, measurement techniques, and analytical decisions necessary to design and interpret gene expression experiments. It is written for life sciences students, laboratory researchers beginning computational analysis, and anyone who needs a practical framework for working with gene expression data.
According to the NCBI Bookshelf, the central dogma of molecular biology provides the fundamental framework: DNA is transcribed into RNA, which is then translated into protein. This flow of information is tightly regulated at multiple steps, and measuring gene expression requires careful experimental design and rigorous quality control. Training resources such as those from EMBL-EBI Training offer structured modules that cover both the biology and the computational methods used in modern gene expression analysis.
At a Glance
The table below summarizes the key components of gene expression and the main techniques used to study them.
| Component | Description | Typical Measurement Method |
|---|---|---|
| Transcription | Synthesis of RNA from a DNA template | RNA sequencing (RNA-seq), quantitative PCR (qPCR), microarrays |
| RNA Processing | Capping, splicing, polyadenylation of pre mRNA | RNA-seq (isoform analysis), long read sequencing |
| Translation | Synthesis of a protein from an mRNA molecule | Ribosome profiling, mass spectrometry |
| Regulation | Control of expression levels via transcription factors, epigenetics, and non coding RNAs | Chromatin immunoprecipitation (ChIP seq), ATAC seq, reporter assays |
The Galaxy Training Network provides interactive tutorials that walk through the computational analysis of each of these measurement types, from raw data to biological interpretation.
Decision Criteria
Before beginning a gene expression study, researchers must choose an approach that fits their biological question, sample type, and resources. The most critical decisions are:
- Targeted versus genome wide measurement. If you already know which genes are relevant (e.g., a small set of candidate genes), qPCR or a targeted gene expression panel is cost effective and precise. If you need to discover new genes or pathways, RNA seq is the standard approach.
- Bulk versus single cell resolution. Bulk RNA seq averages expression over thousands of cells, which is appropriate for most hypothesis driven studies. Single cell RNA seq resolves cell to cell variability but requires more complicated sample preparation and analysis. The Bioconductor project provides dedicated packages for both bulk and single cell analysis, with extensive documentation.
- Quantification method. Microarrays measure relative expression using probe hybridization, RNA seq provides digital counts at base resolution. RNA seq has largely replaced microarrays because it detects novel transcripts, splice variants, and lowly expressed genes with greater dynamic range.
- Replicates and statistical power. Biological replicates are non negotiable. Technical replicates (same sample processed multiple times) are less important than biological variability. For RNA seq, at least three biological replicates per condition are recommended, more are needed for detecting small fold changes or for single cell studies.
The NCBI Sequence Read Archive is a public repository where you can download published RNA seq data to use as controls or to test your analysis pipeline.
Practical Workflow
The following workflow applies to a typical RNA seq experiment, the most common method for genome wide gene expression analysis. Adapt the details for qPCR or microarray experiments accordingly.
1. RNA Extraction and Quality Assessment
Extract total RNA from your samples using a method appropriate for your tissue or cell type. Measure RNA integrity using a Bioanalyzer or TapeStation. The RNA Integrity Number (RIN) should be above 7 for most RNA seq library preparations. Record all metrics, low quality RNA can introduce systematic bias.
2. Library Preparation
Deplete ribosomal RNA or enrich for polyadenylated RNA depending on your goal (e.g., mRNA focused or total RNA including non coding species). Ligate adapters, amplify, and index each sample. Check final library concentration and size distribution.
3. Sequencing
Sequence on an Illumina platform. For bulk RNA seq, aim for 20,30 million reads per sample for gene level quantification. Deeper sequencing may be needed for isoform detection or lowly expressed genes. For single cell experiments, the target is usually reads per cell rather than per sample.
4. Preprocessing
Trim adapter sequences and low quality bases using tools such as Cutadapt or Trimmomatic. The Galaxy Training Network has hands on tutorials for this step. Align reads to a reference genome (e.g., using STAR) or, for non model organisms, perform pseudoalignment with tools like Salmon or Kallisto.
5. Quantification
Summarize aligned reads to gene level counts. For gene level quantification, use featureCounts or HTSeq. For transcript level quantification, use Salmon or Kallisto. Output is a count matrix with rows as genes and columns as samples.
6. Normalization and Differential Expression
Normalize counts to account for sequencing depth and gene length. Common methods are TPM (transcripts per million) or FPKM (fragments per kilobase per million) for relative expression, and DESeq2 or edgeR for statistical testing of differential expression. Apply a false discovery rate (FDR) correction to account for multiple testing.
7. Functional Interpretation
Plot results with volcano plots, heatmaps, and principal component analysis (PCA) to check for batch effects and group separation. Perform gene ontology (GO) enrichment analysis or pathway analysis (e.g., KEGG) to place differentially expressed genes in biological context. The Bioconductor package clusterProfiler is widely used for these analyses.
Quality Checks
Each step in the workflow must include quality checks. Neglecting them is a common source of error.
- RIN and RNA concentration. Record these before library prep. Poor quality RNA leads to uneven coverage and false positives.
- Library fragment size. A tight peak at the expected size (typically 200,500 bp) indicates a successful library. Broad peaks suggest degradation or adapter dimer.
- Sequencing quality scores. The average Phred score should be above 30 for most bases. Use FastQC to generate quality reports. The EMBL-EBI Training materials include a detailed guide to interpreting FastQC outputs.
- Alignment rate. At least 70,80% of reads should map to the reference genome. Low alignment may indicate contamination, poor reference, or sequencing errors.
- PCA on expression data. Replicate samples should cluster together. Outliers may indicate batch effects, sample mix up, or technical failures. Remove or correct batch effects using tools such as ComBat or limma after assessing their impact.
Common Mistakes
Researchers new to gene expression studies often make these errors.
- Using too few biological replicates. Two replicates per group are insufficient for statistical testing. Three is the absolute minimum, four or more are preferable for detecting small but biologically meaningful changes.
- Ignoring batch effects. Samples processed on different days, using different reagent lots, or sequenced on different lanes can produce technical variation. Randomize sample processing and include batch in the statistical model.
- Choosing the wrong normalization method. Comparing raw counts between libraries is invalid. Use proper normalization such as TMM or DESeq2 median of ratios. Do not use RPKM/FPKM for between sample comparisons in differential expression pipelines.
- Overinterpreting small fold changes. Not all genes with a fold change of 1.5 are biologically significant. Focus on statistical significance (adjusted p value < 0.05) and effect size. Validate top hits by an independent method such as qPCR.
- Not checking for sex bias or contamination. In studies using mixed sex samples, include sex specific genes (e.g., XIST, RPS4Y1) in your PCA. If your sample includes multiple species, use a decontamination tool such as FastQ Screen.
Limits of Uncertainty
Gene expression measurements provide a snapshot of RNA abundance, but they do not tell the whole story. Understanding these limits prevents overinterpretation.
- mRNA abundance does not equal protein abundance. Post transcriptional regulation, translation efficiency, and protein turnover can produce large discrepancies between RNA and protein levels. For example, a study of gene expression plasticity in Senegalese sole showed that transcriptional changes did not always correspond to functional protein changes 42443863. Always note that RNA data reflect potential, not actual activity.
- Bulk measurements mask cellular heterogeneity. A doubling of average expression could mean each cell expresses twice as much, or that twice as many cells express the gene. Single cell methods resolve this but introduce sparsity and drop out effects.
- Statistical significance is not biological importance. A gene with a very small p value but a very small fold change may be consistent across replicates but have no functional relevance. Conversely, a highly variable gene with a large fold change may not pass stringent multiple testing correction. Combine effect size and p value in your interpretation.
- Reference genome and annotation quality matter. For non model organisms, transcriptome assemblies may be incomplete. The accuracy of quantification and differential expression depends on the quality of the reference. Use the most current annotation available and consider performing a reference free assembly if necessary.
Frequently Asked Questions
What is the difference between gene expression and transcription?
Gene expression includes all steps from transcription through protein production and regulation. Transcription is specifically the synthesis of RNA from DNA. When researchers measure gene expression by RNA seq or qPCR, they are usually measuring transcript abundance, not the complete expression process.
How do I choose between RNA seq and qPCR?
RNA seq is suitable for genome wide discovery and for measuring thousands of genes simultaneously. qPCR is more appropriate for quantifying a small set of genes (typically fewer than 50) with high accuracy and lower cost. Many studies use RNA seq for discovery and then validate a handful of targets using qPCR.
What is a good number of biological replicates for a differential expression experiment?
For standard bulk RNA seq, three biological replicates per condition is the minimum needed to estimate within group variance. Four to six replicates are recommended when the expected fold changes are small or the sample is highly heterogeneous. Single cell RNA seq typically uses many cells rather than many biological replicates, but at least three biological replicates per condition are still essential for robust inference.
What does a p value of 0.05 mean after multiple testing correction?
After correcting for multiple testing (e.g., using the Benjamini Hochberg method to control the false discovery rate), an adjusted p value of 0.05 means that approximately 5% of the genes declared as differentially expressed are expected to be false positives. This is a balance between sensitivity and specificity.
References and Further Reading
- NCBI Bookshelf , Authoritative textbooks on molecular biology and genomics.
- EMBL-EBI Training , Structured courses on RNA seq analysis, quality control, and data interpretation.
- Galaxy Training Network , Interactive tutorials for bioinformatics workflows, including RNA seq.
- Bioconductor , Open source software and documentation for the analysis of genomic data, with packages for differential expression, normalization, and functional enrichment.
- NCBI Sequence Read Archive , Public repository for raw sequencing data that can be used for testing pipelines or meta analyses.
- Capmatinib and paclitaxel combination: a novel strategy for overcoming challenges in triple-negative breast cancer treatment , An example of a study that used RNA seq to identify differentially expressed genes and pathways.
- Genome-wide analysis and expression profiling of the phenylalanine ammonia-lyase gene family in Chrysanthemum morifolium , A plant study demonstrating gene expression profiling across tissues and stress conditions.
- Single-cell and spatial transcriptomic analysis reveal distinct tumor microenvironment signatures , Example of advanced transcriptomic methods resolving cell type specific expression patterns.