eduKateSG · Why Science?
A high expression level and a large change answer different questions
Find your question below, or follow the complete guide. The reading routes keep the practical tasks and their evidence together.
One gene can be strongly expressed in a sample and still show little change between two groups. Another can have a lower expression level yet show a larger relative difference. That small distinction opens a surprisingly useful science lesson. Expression Atlas helps readers explore expression across biological conditions, but the first question is not which coloured square looks most impressive. It is whether the display answers a baseline question or a comparison question.
This guide explores baseline and differential expression as different reading tasks. It builds on the recently published eduKate guide to Bgee's reference expression without duplicating that resource's role. Here we focus on the defined comparison: which groups, in which direction, under which criteria? All invented genes, plant samples, values and results below are fictional educational data, not Expression Atlas records. The activities require paper, arithmetic and source reading, not biological experimentation.
Resource facts are attributed to EMBL-EBI's official pages. Our study routines and review questions are editorial suggestions, not an official analysis protocol or examination requirement. If you are a parent exploring science enrichment, begin with the numerical examples. If you are a secondary science learner, practise the caption repairs. If you are already reading research, follow the trail from a contrast to its sample descriptions and analysis documentation. There is a useful next step at every level.
Explore the connected science: Bgee and Comparable Reference Gene Expression; Experimental Factor Ontology and Comparable Experimental Variables. For learning and school-choice questions, visit the Education Hub. To reflect on possible future interests, use Careers by Subject Capability. These are exploration routes, not promises of admission, qualifications or employment.
| Question | Record | Do not substitute |
|---|---|---|
| How much in the stated context? | Baseline quantity and unit | Change between groups |
| What differs? | Defined contrast and reference direction | A high baseline value |
| How large is the reported change? | Effect definition and estimate | A significance label alone |
| Which results were shown? | Selected criteria and filters | A claim about omitted results |
| What supports the interpretation? | Design, methods and source trail | A heatmap without context |
Inside this guide
Chapters 1–8 · Ideas and worked examples
- 1. Did You Know? Abundance and change are different questions
- 2. Name both sides of the contrast
- 3. Work through a ratio before using a logarithm
- 4. Read log-two fold change as a defined transformation
- 5. A threshold selects a displayed set; it does not redraw biology
- 6. Keep effect size and statistical evidence side by side
- 7. Replicate values reveal what a summary leaves out
- 8. Check the experiment design before explaining the difference
Chapters 9–16 · Review, learning and next steps
- 9. Do not make separate studies into an undocumented experiment
- 10. An expanded search is a discovery route, not equivalence
- 11. Preserve the gene and feature identity
- 12. Make a contrast card that survives a handover
- 13. A useful bridge between mathematics and biology
- 14. Questions families often ask about expression graphs
- 15. A three-sentence clinic for overconfident conclusions
- 16. Leave with one comparison you can explain honestly
CHAPTER 1 OF 16 · 1. Understand levels, ratios and direction
1. Did You Know? Abundance and change are different questions
Expression Atlas's official introduction describes a resource for expression information across species and biological conditions. Its help separates baseline results from differential results. The first kind concerns expression levels in studied conditions; the second concerns specified comparisons. These are connected views, not interchangeable meanings of the word expression. Start with About Expression Atlas and the official help for the resource's own definitions.
Imagine two fictional plant groups. Gene Beacon has values of one hundred in both groups. Gene Kite has values of five and twenty. Beacon has the larger supplied level in either group, while Kite has the larger supplied ratio between groups. Neither observation makes one gene more important. The exercise has defined two different numerical questions and supplied enough information to answer both without confusing them.
Before reading a heatmap, finish the sentence “I want to know whether ___.” If your blank asks where a gene is expressed, begin with a level-oriented reading. If it asks how expression differs, identify the actual contrast. The reward is immediate: you can choose the appropriate result view instead of trying to extract a change claim from a display that was not answering that question.
CHAPTER 2 OF 16 · 1. Understand levels, ratios and direction
2. Name both sides of the contrast
A comparison needs two identified sides. For this guide's fictional dataset, call them Plant Group A and Plant Group B. State which side is the numerator if a ratio is used. A sentence saying that expression increased is incomplete until the reader knows compared with what. Do not let the familiar name of a condition replace the missing reference group.
Prepare a contrast card with spaces for the question, group descriptions, reference direction and source. Write unknown information as unknown. Then ask a partner to reverse the comparison verbally. If the original card says B relative to A, the reverse is A relative to B. A partner who cannot perform that reversal probably needs a clearer description of the groups before interpreting any colours or numbers.
For real source reading, use the experiment's documented comparison rather than reconstructing it from a shortened plot label. The official help explains that comparison information includes experimental variables and sample characteristics. Our recommended notebook records that context alongside the result. This simple habit makes it harder to turn “higher in the stated comparison” into “high everywhere,” or to accidentally reverse a conclusion while copying a chart into a presentation.
CHAPTER 3 OF 16 · 1. Understand levels, ratios and direction
3. Work through a ratio before using a logarithm
Use these fictional supplied values: Group A has ten units and Group B has forty units under the exercise's defined measurement scale. The B-to-A ratio is forty divided by ten, which equals four. The A-to-B ratio is one quarter. Both calculations describe the same pair of values, but in opposite directions. Neither should be described as a four-unit difference: the arithmetic difference is thirty units.
Now compare a second fictional pair, one hundred and one hundred and thirty units. Its difference is also thirty, but its B-to-A ratio is one point three. The same additive difference does not imply the same relative change. Ask learners to explain this with a drawing of bars before introducing any specialist terminology. The diagram should label both the quantities and the comparison direction.
This is an arithmetic exercise, not an estimate of a biological effect or an attempt to reproduce Expression Atlas's analysis. Its purpose is to make the number's meaning recoverable. When reading an actual result, preserve the defined quantity and analysis method rather than assuming that a ratio of two visible summary numbers is the reported statistical estimate. Clear elementary reasoning and careful source reading should support each other.
CHAPTER 4 OF 16 · 1. Understand levels, ratios and direction
4. Read log-two fold change as a defined transformation
For positive values in our fictional ratio exercise, log-two of four equals two because two raised to the power two equals four. Log-two of one quarter equals minus two. Log-two of one equals zero. These examples show how equal-magnitude positive and negative log ratios can represent reciprocal comparisons. They do not establish a universal rule for handling zero measurements, uncertainty or normalisation in research data.
Write the ratio beside its log-two value until the relationship feels familiar. A learner who sees plus two should be able to explain the fourfold ratio under this exercise's definition, not a two-unit increase. A learner who sees zero should identify equal supplied positive values, not absence of expression. The phrase zero change and the phrase zero expression answer different questions.
Expression Atlas's differential help identifies log-two fold change among the reported quantities. For a real dataset, read its analysis documentation before interpreting the estimate. Do not invent a correction for zero values or calculate a new result and present it as the resource's output. The safe educational achievement is understanding the transformation well enough to ask a precise question about the reported number and the contrast that gives its sign meaning.
CHAPTER 5 OF 16 · 2. Read criteria, variation and study design
5. A threshold selects a displayed set; it does not redraw biology
The Expression Atlas help describes controls for changing differential-expression display criteria, including log-two fold change and adjusted p-value. Consult the current interface and documentation for the actual settings rather than assuming a remembered default applies to every view. A filtering decision belongs in the evidence note because it affects which results the reader sees.
For a fictional exercise, supply three already-calculated results with effect values of zero point nine, one point one and two point zero. Define a teaching filter that displays only values greater than one. Two results appear and one does not. Changing the filter to greater than one point five leaves only the last. Nothing about the supplied underlying values changed when the filter changed.
Ask learners to repair “The hidden item has no effect.” A better sentence says that it did not meet the selected display rule in the exercise. Then ask what else would be needed to judge the result scientifically. The filter alone is not that judgement. This is a practical connection to school science: whenever a graph or table shows only a selected subset, the selection rule should travel with the conclusion rather than disappear behind a clean presentation.
CHAPTER 6 OF 16 · 2. Read criteria, variation and study design
6. Keep effect size and statistical evidence side by side
The official differential help presents fold-change information alongside adjusted p-values. We will not turn this introductory article into a statistical testing manual. The useful reading task is to keep the reported effect quantity distinct from the reported statistical evidence and to consult the stated method. Do not convert an adjusted p-value into a probability that the whole scientific explanation is true.
In a fictional table, Result One has a larger effect estimate but is marked as not meeting the supplied statistical rule. Result Two has a smaller effect estimate and meets that rule. Ask which result has the larger effect and which meets the rule. These questions have separate answers. The exercise supplies the classifications; students are not deriving p-values from a few invented measurements.
The final explanation should report the quantities in their defined roles rather than rank everything with the vague word stronger. If stronger means larger estimated difference, say that. If it means meeting a stated criterion, name the criterion. If the task concerns practical relevance, that needs its own argument. This precision makes the result easier to discuss without asking a learner to pretend that one number answers every scientific question.
CHAPTER 7 OF 16 · 2. Read criteria, variation and study design
7. Replicate values reveal what a summary leaves out
The Expression Atlas help describes a baseline boxplot view for variation among biological replicates. That is a reason to inspect more than the central heatmap value when the question concerns variation. It is not a guarantee that every dataset has the same replicate structure or that a particular plot alone is sufficient for analysis. Read the experiment context before interpreting the display.
Try two fictional groups with three supplied values each. The first is eight, ten and twelve; the second is two, ten and eighteen. Both arithmetic means are ten. Their ranges differ: four in the first group and sixteen in the second. The same mean therefore does not describe the same spread in these lists. No significance test or biological population estimate is implied by this small example.
Draw the six points on a number line before showing the means. Ask which differences disappear when the points are replaced by one number per group. This gives learners a concrete reason to read variation rather than treat it as decorative statistical information. For real reporting, distinguish a copied summary from an independent calculation and explain which samples it summarises. More decimal places in the mean cannot restore information about the individual observations that was not reported.
CHAPTER 8 OF 16 · 2. Read criteria, variation and study design
8. Check the experiment design before explaining the difference
A biological comparison can involve several described characteristics, not only the label highlighted in a headline. Our recommended review reads the group definitions before deciding which interpretation is justified. The official help provides a route to experiment-design information and analysis documentation. Those sources are more appropriate than guessing missing context from the colour of a result cell.
Use fictional plant cards with group labels, age labels and a batch label. In one version, all Group A cards come from the first batch and all Group B cards from the second. The supplied description does not let a learner separately identify a group explanation and a batch explanation from these labels alone. This exercise is about recognising the ambiguity, not correcting a real experiment or prescribing a design.
Write the observed contrast and the proposed explanation on different lines. A student may still describe the supplied difference accurately while leaving its explanation unresolved. That is a useful outcome. It identifies what the source would need to establish for a stronger claim. For a parent or tutor, the key feedback is concrete: ask the learner which documented fact separates the explanation they prefer from an alternative they have not yet ruled out.
CHAPTER 9 OF 16 · 3. Trace the contrast and its source
9. Do not make separate studies into an undocumented experiment
Finding one value in one study and another value in a different study can inspire a comparison question. It does not automatically create a controlled contrast between the studies. The reader needs to review the measurements, sample descriptions and analysis context. This article's editorial recommendation is to keep within-study reported comparisons separate from a newly proposed cross-study comparison.
For a fictional reading task, one study reports a value for Group A and another reports a value for Group B. Their units have the same name, but the exercise supplies no evidence about processing equivalence or group matching. Ask learners to write what was actually retrieved: two reported values from two sources. Then write the comparison they would like to make and list the missing justification.
The recently published Bgee guide explores a different reference-comparison setting. Do not transfer a resource-specific score definition or curation rule from that article to Expression Atlas. The broader skill is to identify which resource and quantity you are using. Shared interest in gene expression does not make every table, scale or comparison interchangeable, and a source-aware explanation should let the reader see the difference.
CHAPTER 10 OF 16 · 3. Trace the contrast and its source
10. An expanded search is a discovery route, not equivalence
Expression Atlas's introduction explains its use of Experimental Factor Ontology mappings to support richer searches. A search may therefore retrieve related condition terms rather than only a literal phrase. This can be helpful for discovery, but the reader should inspect the actual experiment descriptions before combining the returned results. Read the existing Experimental Factor Ontology guide for the broader vocabulary question.
For a fictional library exercise, a search for leaf includes records labelled broad leaf and leaf subdivision. Those results may be related, but they do not necessarily concern identical samples or the same contrast. Ask learners to put the original query, returned label and accepted use on separate cards. This makes the path from discovery to selection explicit.
The final report should explain why a retrieved study was relevant to the question, not simply state that the search returned it. If the retrieved condition is broader or narrower than intended, preserve that fact. A useful next step may be to narrow the question or review another source. Good search tools help us find candidates; careful science reading decides which candidate supports the claim we are preparing to make.
CHAPTER 11 OF 16 · 3. Trace the contrast and its source
11. Preserve the gene and feature identity
A gene symbol helps readers recognise a result, but a source review should preserve the supplied identifier and organism as well. Expression Atlas's help also notes that microarray results can involve multiple design elements associated with a gene. Do not silently treat every row as an independent gene or every repeated label as an accidental duplicate. Read the particular experiment's feature definitions.
Create a fictional table with two local probe labels assigned to one invented gene. Ask how many rows and how many represented genes the table contains. There are two rows and one gene under the supplied mapping. That simple count changes if the mapping is removed or disputed. Keep the row-to-gene relationship rather than relying on the spreadsheet's row total to answer the biological counting question.
Our suggested report has a sentence explaining what each row represents. This is useful even when the table is small. It prevents a reader from confusing genes, transcripts, features, samples and comparisons simply because all appear as rows somewhere in a workflow. The immediate learning gain is a more faithful count and a clearer caption, not a claim that the learner has independently resolved all identifier mappings used in the underlying study.
CHAPTER 12 OF 16 · 3. Trace the contrast and its source
12. Make a contrast card that survives a handover
Bring the reading together on one page: experiment identifier, organism, gene or feature reference, group definitions, comparison direction, reported quantity, selected criteria, source links and unresolved limitation. These are recommended educational fields, not an official submission template. Include only information you actually established. A blank marked unknown is more useful than a polished but invented completion.
Give the card to a partner who has not seen the heatmap. Ask them to explain what plus two means in the fictional log-two example, which group is the reference and what the display filter did. If they cannot answer, revise the card before adding another chart. This tests whether the explanation is portable rather than dependent on the author standing beside it.
For real source reading, include the retrieval date and relevant analysis-documentation link. Record whether the result was copied from the resource or recalculated in a separate educational exercise. A partner should be able to follow the observation without mistaking an illustrative calculation for a downloaded statistical result. The handover is successful when another reader can identify both the supported comparison and the claim you deliberately left open.
CHAPTER 13 OF 16 · 4. Practise a clear scientific explanation
13. A useful bridge between mathematics and biology
This topic shows why science learning benefits from understanding ratios, powers, graphs and variation. The fictional examples can be taught with ordinary arithmetic before specialist vocabulary is introduced. That makes the lesson accessible without implying that every school assessment requires Expression Atlas or differential-expression analysis. Treat the resource as optional enrichment, not an additional examination burden.
For a younger learner, compare two positive quantities and explain the difference between adding thirty and multiplying by four. A secondary learner can reverse a ratio and describe the effect on direction. A more confident learner can connect powers of two to the fictional log transformation and inspect a source's quantity definition. Choose the task that fits current readiness rather than adding technical language simply to make the activity look advanced.
If you are exploring science tuition in Singapore, ask how a tutor helps a learner connect an unfamiliar graph to a familiar mathematical relationship. Look for a worked explanation and a chance to correct a misconception. No resource visit guarantees a grade improvement. The Science Learning Hub offers a broader starting point when the specialist example needs to be connected back to everyday learning goals.
CHAPTER 14 OF 16 · 4. Practise a clear scientific explanation
14. Questions families often ask about expression graphs
Does a larger baseline value imply a larger change? Not in the supplied arithmetic examples: level and contrast are separate quantities. Does log-two fold change of zero mean no expression? No; under the positive-value ratio definition used here it means equal compared values. Can a colour be read without its legend? A responsible explanation needs the relevant scale, quantity and selected comparison.
Can the result prove a treatment works? This article does not establish any treatment outcome. A differential-expression result concerns its documented comparison, and wider causal or practical claims require suitable additional evidence. Can a parent use the activity without specialist software? Yes: the fictional arithmetic, caption review and contrast cards are designed for paper-based learning. They are editorial suggestions, not a substitute for professional analysis.
Where should school-choice and qualification questions go? Use the Education Hub to organise those wider questions and current official sources for specific requirements. Keep enjoyment of this enrichment task separate from claims about admissions advantage. The useful immediate outcome is clearer reasoning: a learner who can state what two groups were compared and what the number means has made meaningful progress without needing to choose a specialist pathway today.
CHAPTER 15 OF 16 · 4. Practise a clear scientific explanation
15. A three-sentence clinic for overconfident conclusions
First sentence: “The high-expression gene changed most.” Repair it by identifying whether the source supplies a level or a change estimate, then reporting the appropriate quantity. Second sentence: “The result disappeared, so the gene stopped being expressed.” Repair it by explaining the selected filter and distinguishing a hidden result from an observed change. Third sentence: “Plus two means two more molecules.” Replace that with the documented quantity definition and comparison direction.
For each repair, preserve the interesting observation. The aim is not to make every sentence hesitant, but to make it accurate. Ask a partner to underline the evidence-bearing phrase and circle the phrase that added an unsupported claim. This makes feedback specific enough to use. If the correction still relies on “be careful,” name the actual ambiguity instead.
Finish by writing your own fourth example using the fictional data. Supply an answer explaining why the revised sentence is better and what remains unresolved. This turns source literacy into an active skill. It also gives parents and tutors a constructive way to respond to confident mistakes: keep the curiosity, identify the numerical or contextual shift, and rebuild the sentence so that another reader can see exactly what supports it.
CHAPTER 16 OF 16 · 4. Practise a clear scientific explanation
16. Leave with one comparison you can explain honestly
Your final mini-report should answer the original question using identified groups, a stated direction and a defined quantity. Include the source and selection criteria, and separate a reported comparison from a new interpretation. If a necessary detail could not be established, make the open question specific. A concise, inspectable conclusion is more useful than a dramatic claim whose reference group has vanished.
This work can also reveal interests worth exploring: mathematical explanation, biological questions, source organisation or data visualisation. Use Careers by Subject Capability for a broader reflection, not a promise of a qualification or job. Choosing one skill to practise next is enough. You might improve a graph caption or explain a reciprocal ratio to a friend.
Why science? Because the meaning of a result depends on the question it answers. Expression Atlas gives us a practical setting for seeing how abundance, change, variation and comparison context fit together. Start with one contrast, keep both sides visible and explain the number before extending the conclusion. That small habit makes scientific evidence easier to understand, share and question—and keeps discovery grounded in something another learner can follow.
Sources and further reading
Source review: 17 October 2026. Resource facts are attributed to the primary documentation below; the teaching activities, review questions and suggested study routines are eduKate editorial guidance. Fictional examples do not report research findings. Interface details and records may change: check the linked source before specialist reuse.
- EMBL-EBI: About Expression Atlas
- Expression Atlas: official help pages
- EMBL-EBI training: baseline and differential contexts
- Expression Atlas: download documentation
This article is an educational introduction, not a laboratory protocol, medical recommendation or professional certification. It does not claim affiliation with the named projects. Start with the Science Learning Hub when you want a broader learning route.