eduKateSG · Why Science?
An orthology call becomes understandable when the gene tree, species tree, duplication history and release stay connected
Follow Ensembl Compara from gene family and tree to speciation or duplication node, relationship cardinality and quality scores before using a cross-species gene claim.
When two genes in different species look related, it is tempting to call them equivalents. Evolutionary history is more precise. Genes can diverge when species split, duplicate within a lineage, disappear from sampled genomes or acquire complex many-to-many relationships. Ensembl Compara models that history with gene trees reconciled against species trees.
The result is useful because it makes an orthology claim inspectable. Readers can trace genes to a node, identify a speciation or duplication event, examine relationship cardinality and record the Ensembl release. This guide turns comparative genomics into a careful science-learning exercise rather than a one-word lookup.
Use a small gene family first. Keep the browser tree beside a worksheet with columns for gene ID, species, tree ID, node type, homology type, cardinality, quality fields and release. The worksheet slows the jump from a familiar symbol to a sweeping functional claim and makes each inference visible to a partner.
Tree diagrams compress many decisions. Leaves depend on genome annotations, alignments and family membership; internal nodes reflect an inferred history; labels summarise pipeline rules. None of this makes the result arbitrary. It means the result is a model whose inputs and method can be named, inspected and compared with other evidence.
Read the display from the selected leaves inward. First confirm both gene records, then locate their common node, then interpret the node using the reconciled species history. Only after that should you copy a pairwise relationship label. This order prevents a convenient table cell from becoming detached from the tree that explains it.
Release archives are central to reproducibility. A current browser may reflect newer gene annotations than a paper or classroom worksheet. When the evidence was collected from the June 2026 release 116 archive, retain that archive URL and release number in the citation. A later comparison can then identify what actually changed.
Comparative genomics also teaches a broader lesson about similarity. Sequence resemblance, evolutionary relationship, shared function and comparable phenotype are connected possibilities, but they are not the same observation. Each deserves its own source and verb. Separating them produces a more credible synthesis and exposes interesting questions that a single word such as “equivalent” would hide.
Students can share the reasoning roles: one follows the gene tree, one checks the species tree and release, and one searches for independent functional evidence. When the three accounts agree, the claim gains a transparent chain. When they differ, the discrepancy becomes a learning opportunity rather than something to conceal.
Before exporting a table, decide whether the unit of analysis is a gene, transcript, protein or pairwise relationship. A single gene can participate in several reported relationships, so row count is not automatically gene count. Label the unit beside every total and avoid deduplicating rows until the relationship structure is understood.
Tree graphics also require accessible explanation. Use species names and gene IDs in accompanying text, describe the selected node and avoid relying only on branch colour. If a large tree is pruned for clarity, state which leaves were hidden and keep the complete tree identifier so the full context remains retrievable.
Negative findings need careful search boundaries. Failure to retrieve an orthologue from one species under one release and method does not prove that no related gene exists. Record the species assembly, analysis type, filters and release, then decide whether another database or sequence search is needed for the question.
For project review, choose one sentence and trace every noun. The gene should resolve to a stable record, the species to the intended assembly, the tree to a release-specific identifier and the relationship to a labelled node or pairwise output. This noun audit often catches problems before sophisticated evolutionary language hides them.
A simple classroom investigation can compare one selected gene across three species. Students should sketch the relevant subtree, list relationship cardinalities and add one independent functional source per species. The result is a compact evidence package rather than a claim that all related genes are interchangeable.
When results change between releases, compare inputs before declaring that evolutionary history changed. A revised gene model, new assembly, altered species sampling or updated pipeline may change the inferred tree. Version comparison is therefore both biological and computational, and the report should say which layer shifted.
Keep uncertainty proportional to the decision. A classroom comparison may need only a clear release, relationship type and limitation; a research claim may require alignments, alternative trees and sensitivity analyses. More detail is useful when it answers a risk in the interpretation, not when it merely makes the report look technical.
This layered approach gives every reader a way to verify the core claim and a clear route toward deeper analysis.
Chapter index
Chapters 1–9: from gene identity to reconciled trees
- 1. Anchor both genes and the release
- 2. Separate gene tree and species tree
- 3. Define orthology at a speciation node
- 4. Define paralogy at a duplication node
- 5. Read one-to-one and many-to-many relationships
- 6. Trace the relevant node
- 7. Treat missing leaves as questions
- 8. Preserve gene-tree stable identifiers
- 9. Inspect identity, coverage and quality fields
Chapters 10–17: from relationship types and scores to careful comparison
- 10. Add gene order and whole-genome context
- 11. Interpret high-confidence labels
- 12. Compare resources method by method
- 13. Keep analysis families distinct
- 14. Select a reproducible access route
- 15. Use evolutionary verbs accurately
- 16. Join function only after relationship
- 17. Connect evolution, algorithms and study choices
Begin with stable Ensembl gene identifiers, species names and the Ensembl release. Gene symbols are convenient for conversation, but they can be shared, revised or interpreted differently across species. Stable identifiers provide a much stronger audit trail.
Try this
Create a two-row identity card with gene ID, symbol, species, assembly and transcript or protein choice where relevant. Check that the records are current in the same release before opening a comparative view.
Release context matters because genome annotations and tree membership can change. A claim retrieved from the June 2026 release 116 archive should be labelled that way rather than silently presented as timeless.
Evidence checkpoint
Record Ensembl gene identifiers, species and release before opening the tree. Symbols alone cannot anchor a comparative claim.
A species tree represents the branching history of species. A gene tree represents inferred relationships among genes or their products. Ensembl Compara reconciles the two so internal nodes can be interpreted as likely speciation or duplication events.
Try this
Draw a simple species tree for three organisms, then place two gene copies in one lineage. Compare it with the resulting gene tree. This exercise shows why one species can contribute several related leaves and why those leaves need not have identical histories.
Never describe the gene tree as a direct measurement of function. It is a model inferred from sequence and pipeline methods, interpreted in the context of species relationships and genome annotation.
Evidence checkpoint
Distinguish the gene tree from the species tree. Reconciliation uses both to infer where duplication and speciation events occurred.
Orthologues are genes whose relationship is traced through a speciation event in the reconciled tree. The definition concerns evolutionary history. It does not guarantee identical expression, phenotype, biochemical activity or clinical relevance.
Try this
Select two leaves from different species and trace their branches backward to the relevant node. Record how the interface labels that node and the reported homology type. The visible path is stronger evidence than copying “orthologue” from a summary list.
Write the bounded statement first: “Compara release 116 infers an orthology relationship between these gene records.” Add functional similarity only after consulting independent evidence appropriate to that claim.
Evidence checkpoint
Treat an orthologue as a relationship inferred across a speciation event, not a guarantee of identical function.
Paralogues arise through a gene duplication relationship. The duplication may predate or follow a species split, which changes how copies are distributed across lineages. The timing is therefore part of the interpretation.
Try this
Find a duplication node and note the descendant genes and species. Ask whether both copies persist in each lineage or whether gains, losses and annotation coverage produce a more uneven pattern.
Paralogy also does not specify function by itself. Duplicated genes may retain, divide or change roles, but those possibilities need expression, phenotype, biochemical or other functional evidence.
Evidence checkpoint
Treat a paralogue as a relationship associated with gene duplication, and keep the timing relative to species divergence visible.
CHAPTER 5 OF 17 · Read speciation and duplication
5. Read one-to-one and many-to-many relationships
One-to-one, one-to-many and many-to-many labels describe the cardinality of inferred orthology relationships between two lineages. They are structural descriptions of evolutionary history, not league tables of which gene is most important.
Try this
For a one-to-many example, place the single gene from one species opposite the duplicated genes from the other. Trace the duplication relative to speciation. This makes the label understandable without memorising it.
A one-to-one label can simplify comparison, but it still does not certify functional equivalence. A many-to-many relationship may be harder to analyse, yet it can contain biologically important lineage-specific expansion.
Evidence checkpoint
Read one-to-one, one-to-many and many-to-many cardinality as relationship structure, not a ranking of biological importance.
Large gene trees can contain many leaves and several duplications. Start from the selected gene, follow its branch toward the comparison gene and stop at their most recent common node. Do not interpret labels from a distant part of the tree.
Try this
Record the node type, descendant leaves and any confidence information shown. If the display collapses branches, expand only the path needed for the claim and keep a note of hidden groups.
This method turns tree reading into an explicit reasoning sequence. Another student can follow the same leaves and confirm whether the relationship is supported at the same node.
Evidence checkpoint
Follow tree branches from the selected gene to the relevant node before reading labels at distant parts of a large family.
CHAPTER 7 OF 17 · Read speciation and duplication
7. Treat missing leaves as questions
A species may lack a leaf in a gene tree because of genuine gene loss, incomplete genome sequence, annotation gaps, filtering or the boundaries of the family model. The tree alone may not distinguish every explanation.
Try this
Check the species and gene records, genome assembly and relevant method documentation before writing “the species lacks this gene.” Search for unplaced or alternative annotations only if the project requires that level of investigation.
Use “no corresponding leaf was present in this Compara tree and release” as the initial observation. A claim of biological absence requires additional evidence that the genome and analysis had a fair chance to find the gene.
Evidence checkpoint
Keep gene gains, losses and incomplete coverage as model possibilities. A missing leaf is not automatically biological absence.
CHAPTER 8 OF 17 · Read speciation and duplication
8. Preserve gene-tree stable identifiers
Ensembl provides stable identifiers for gene trees, but stability is conditional. Membership and topology can change between releases, and sufficiently large changes may lead to a different identifier or version context.
Try this
Save the tree stable ID with the Ensembl release and the selected gene IDs. When revisiting a project, compare membership and the relevant node rather than assuming an unchanged label means an unchanged tree.
This is an important provenance lesson: stable does not mean eternal or context-free. It means the resource supplies a managed identity that becomes useful when version and release are also recorded.
Evidence checkpoint
Use tree stable identifiers with release context because membership changes can affect whether an identifier persists across releases.
CHAPTER 9 OF 17 · Inspect confidence and releases
9. Inspect identity, coverage and quality fields
Pairwise displays may include percentage identity, coverage and other quality information. Each field answers a particular question and depends on the sequences and alignment used. None should become a universal confidence score on its own.
Try this
Compare identity in both directions when coverage differs. A short protein matching part of a longer protein can yield a high local identity while leaving much of one sequence outside the comparison.
Record the field name and value exactly, then explain its limitation. The tree relationship, alignment quality and annotation context work together; a single percentage cannot replace them.
Evidence checkpoint
Inspect percentage identity, coverage and available quality scores without turning one metric into universal confidence.
CHAPTER 10 OF 17 · Inspect confidence and releases
10. Add gene order and whole-genome context
Gene order conservation and whole-genome alignment evidence can provide independent context for orthology relationships. They ask whether surrounding genomic organisation or aligned regions support the pairing in addition to the gene tree.
Try this
Inspect the Gene Order Conservation score or whole-genome alignment score when supplied and read the release-specific definition. Treat the scores as additional evidence rather than a substitute for the tree inference.
Apparent disagreement can be informative. Rearrangement, assembly quality, rapid evolution or annotation differences may weaken one line of context without automatically invalidating every other line.
Evidence checkpoint
Gene order conservation and whole-genome alignment scores provide independent context; they do not replace the tree inference itself.
CHAPTER 11 OF 17 · Inspect confidence and releases
11. Interpret high-confidence labels
High-confidence orthology labels are produced under criteria defined by the Compara pipeline and release. They summarise selected evidence and thresholds; they do not convert an inference into direct functional measurement.
Try this
Open the quality-control documentation for the same release and identify which metrics contribute to the label. If you filter a dataset by confidence, record the rule and count relationships removed.
Avoid writing simply “confirmed orthologue.” Instead, state that Ensembl labels the relationship high confidence under its documented method, then preserve the underlying homology type and scores.
Evidence checkpoint
Check high-confidence labels and the method definition used in the relevant Ensembl release before filtering relationships.
CHAPTER 12 OF 17 · Inspect confidence and releases
12. Compare resources method by method
Orthology resources can disagree because they use different genome releases, gene sets, sequence choices, tree methods, species sampling and confidence rules. Counting disagreement without aligning these inputs can misrepresent both resources.
Try this
Build a comparison sheet with species, gene versions, resource release, method family, relationship label and retrieval date. Restrict the first comparison to matched objects and explain any records that could not be aligned.
Differences are not automatically errors. They can reveal sensitivity to annotation or modelling assumptions and suggest which branch, sequence or genome region needs closer inspection.
Evidence checkpoint
When comparing with another orthology resource, align species, gene versions, methods and release dates before counting disagreements.
Compara includes protein trees, ncRNA trees, pairwise homologies, multiple alignments and other comparative analyses. Their input objects and methods differ. A statement about a protein-coding gene tree should not be transferred casually to an ncRNA analysis.
Try this
Identify the analysis type shown in the browser or returned by an API. Record whether the relationship is protein-based, nucleotide-based or derived from another comparative pipeline.
This distinction helps students see that “comparative genomics” is a family of methods. Choosing the correct method for the biological object is part of evidence quality.
Evidence checkpoint
Separate coding-gene protein trees from ncRNA trees and other Compara analyses; methods differ across object types.
The Ensembl browser is excellent for visual exploration. BioMart, REST endpoints or programmatic APIs can support larger or repeatable retrievals. The best route is the one that preserves the identifiers, relationship fields and release needed for the question.
Try this
For a small project, save the relevant table and a screenshot of the node path. For a script, record endpoint, parameters, response fields and date. Use an archive server when a fixed historical release is required.
Compare one programmatic result with the browser view before scaling. Field names and defaults can affect interpretation, and validation against a known case catches silent assumptions early.
Evidence checkpoint
Use the browser, BioMart, REST or API route that preserves the fields needed for reproducibility, and record the route used.
Compara infers, reconciles, classifies and scores evolutionary relationships. It does not directly measure gene function, disease risk or phenotype. Those claims require their own evidence sources.
Try this
Rewrite “this gene has the same function in both species” as “Compara infers these genes to be orthologues in release 116; the following functional records provide separate evidence for…” The two-part statement preserves both the evolutionary result and the functional test.
Precise verbs make cross-species reasoning stronger. They show exactly where an inference enters and prevent a convenient label from carrying more biological meaning than its method supports.
Evidence checkpoint
Write that Compara infers, reconciles, classifies or scores relationships. Do not claim that a tree directly measures function.
CHAPTER 16 OF 17 · Build a comparative claim
16. Join function only after relationship
Once the evolutionary relationship is clear, connect each gene to reviewed functional, expression or phenotype evidence. Keep species-specific identifiers and citations so evidence from one organism is not silently copied to another.
Try this
The Gene Ontology guide helps readers inspect annotation evidence, while the Alliance of Genome Resources guide supports careful cross-species record linking.
Use orthology to choose a comparison candidate, then ask whether experimental conditions, tissues and phenotypes are comparable. Evolutionary relatedness guides the question; it does not answer every functional part.
Evidence checkpoint
Link functional or phenotype claims to their own sources after orthology identifies a comparison candidate.
CHAPTER 17 OF 17 · Build a comparative claim
17. Connect evolution, algorithms and study choices
Compara sits at the meeting point of evolution, statistics, algorithms and genome annotation. A student might pursue biology, computer science, medicine or data curation, but a single database investigation cannot promise a course place or career.
Try this
Continue through the Science Learning Hub for concepts and the Education Hub for broader planning. The next capability might be tree reading, sequence alignment, programming or evidence writing.
Did You Know? “One-to-one orthologue” describes an inferred history between lineages, not a perfect biological copy. Recognising that distinction is a powerful step from memorising gene names to reasoning about evolution. A bounded project can compare one gene across three species while recording the exact tree, relationship fields and release.
Comparative relationships can help prioritise conserved genes or residues, while variant consequence prediction asks a separate question about a sequence change and transcript. The Ensembl VEP guide is a useful next step once identity and evolutionary context are secure.
Evidence checkpoint
Connect evolutionary reasoning, algorithms and data provenance to further study without promising a specific course or career result.
Contents · Previous chapter · Continue at the Science Learning Hub
A final gene-tree reading routine
Before leaving an Ensembl Compara view, capture both gene identifiers, species, tree stable identifier when shown, homology type, cardinality, quality fields and the Ensembl release. Trace the selected leaves to the node that supports the relationship. This makes the orthology statement reconstructable instead of relying on a single label copied from a summary table.
Then open a functional or phenotype source only after the evolutionary relationship is clear. Orthology helps choose a comparison; it does not carry every annotation across species. A careful report says what the tree inference supports and separately cites the evidence used to compare function, expression or phenotype.
| Reading object | What it can support | Question to keep visible |
|---|---|---|
| Gene tree | A model of relationships among genes | Which leaves and internal node support the comparison? |
| Species tree | The reference branching history of species | How was the gene tree reconciled with species divergence? |
| Orthology or paralogy call | A pairwise relationship inferred from speciation or duplication | What homology type and cardinality are reported? |
| Quality score or label | Independent context for an inferred relationship | Which metric, threshold and release produced the label? |
Sources and next reading
Substantive resource claims in this article were checked against the primary pages below. A database record is a structured route into evidence; it does not by itself prove a new biological conclusion. Recheck the live record, evidence labels and release context before using it in assessed work or research.
Primary sources
- Ensembl Compara comparative genomics overview, release 116 archive — accessed 24 October 2026.
- Ensembl Compara protein-tree method — accessed 24 October 2026.
- Ensembl Compara homology types — accessed 24 October 2026.
- Ensembl Compara gene-tree stable identifiers — accessed 24 October 2026.
- Ensembl Compara orthology quality controls — accessed 24 October 2026.