Background
Differential gene-expression analysis can identify genes whose expression changes between experimental conditions, but a long list of individual genes can be difficult to interpret biologically.
This project examined estradiol-responsive gene expression in MCF7 breast cancer cells and used functional enrichment analysis to move from individual differentially expressed genes to higher-level biological processes.
The central question was: which cellular functions and molecular processes are disproportionately represented among genes whose expression changes following estradiol treatment?
Research Goals
- Analyse a set of differentially expressed genes (DEGs) associated with estradiol treatment in MCF7 cells.
- Map the DEG set to Gene Ontology terms and biological functions.
- Identify statistically enriched biological processes while accounting for the large number of simultaneous hypothesis tests.
- Interpret enrichment patterns related to hormone response, cell-cycle regulation, and DNA replication.
- Distinguish statistical enrichment from stronger claims about causal biological mechanisms.
Dataset
The analysis used differentially expressed genes derived from publicly available transcriptomic data comparing control and estradiol-treated MCF7 cells.
Public biological resources were used to connect gene identifiers with functional annotations and Gene Ontology terms.
Methodology
Functional Enrichment
Used g:Profiler to test whether particular biological functions were overrepresented in the DEG set.
Examined enriched terms from Gene Ontology categories including:
- Biological Process (BP)
- Molecular Function (MF)
Used biological annotation resources including Ensembl BioMart and public gene-expression databases.
Multiple Testing
Functional enrichment involves testing many ontology terms simultaneously, creating a substantial risk of false-positive findings.
I therefore incorporated multiple-testing correction when evaluating enrichment results, reducing the likelihood of interpreting terms that appeared significant only because many hypotheses were tested simultaneously.
Biological Interpretation
The statistically supported terms were grouped and interpreted according to their biological relevance, with particular attention to:
- Cell-cycle regulation
- DNA replication
- Hormone-responsive processes
- Estrogen-related transcriptional activity
The analysis treated these findings as associations present in the DEG set, rather than direct experimental evidence that individual pathways were causally activated by estradiol.
Results
The enrichment analysis identified several biological processes that were disproportionately represented among estradiol-responsive genes.
Prominent patterns included terms related to:
- Cell-cycle regulation
- DNA replication
- Response to hormone stimulus
- Other hormone-responsive cellular processes
These patterns provided a higher-level biological interpretation of the differential-expression results.
Rather than examining genes individually, enrichment analysis showed that the transcriptional changes could be organised into coherent groups of related biological functions.
A second important result was methodological: interpretation depended not only on individual enrichment scores but also on accounting for the large number of hypotheses tested simultaneously. Multiple-testing correction therefore formed an important part of determining which enrichment signals were sufficiently supported for further interpretation.
Contributions
- Translated a large differential gene-expression set into interpretable biological functions.
- Applied Gene Ontology enrichment analysis using g:Profiler.
- Incorporated multiple-testing correction to reduce false discoveries arising from simultaneous hypothesis testing.
- Identified enrichment patterns associated with cell-cycle regulation, DNA replication, and hormone response.
- Combined computational results with biological annotation to interpret transcriptomic changes at a systems level.
- Explicitly distinguished statistical enrichment from causal biological conclusions.
Tools and Technologies
- Databases and Resources: GEO, Ensembl BioMart
- Functional Analysis: g:Profiler, Gene Ontology
- Programming: Python
- Methods: Differential-expression interpretation, functional enrichment, multiple-testing correction
Impact
This project demonstrated how computational analysis can transform a large collection of molecular measurements into a smaller set of biologically interpretable hypotheses.
It also highlighted an important limitation of data-driven analysis: statistical enrichment identifies associations within the analysed gene set, but does not by itself establish a causal biological mechanism.
The project provided practical experience in multiple-testing control, biological interpretation, and distinguishing what a quantitative result supports from conclusions that extend beyond the evidence.