Biology · Genomics, Omics & Bioinformatics
Transcriptomics, proteomics, metabolomics and integrated measurements
An invented RNA reading doubles while a corresponding protein reading stays the same and a metabolite reading halves. These changes can be recorded together without adding the three numbers or assuming that one change caused the others. A second table repeats measurements from four imagined biological units; eight readings still represent four units.
Gene Expression: RNA & Proteins explains how sequence, RNA and protein are related. Phylogeny, Classification & the Tree of Life introduces ancestry and the interpretation of trees. Research Design, Data & Scientific Practice explains sampling, independent units and the limits of comparisons. Their English chapters are available in this working manuscript; publication remains pending. The examples here use letter strings, inclusive coordinates, averages, fractions and ordinary graph reading. Each convention is introduced before it is used.
Back to the Biology mapBy the end, you can
Connect different molecular measurement layers while respecting their scales, timing, replication and limits as evidence for mechanism.
Each measurement layer has its own question
Transcriptomics studies RNA patterns, proteomics studies proteins, and metabolomics studies small-molecule metabolites. A measurement can describe an amount, concentration, relative signal or another defined property. These quantities require their own units, timing and measurement interpretation. A large value on one instrument's scale is not automatically a larger biological amount than a value on another scale.
RNA production, protein production and molecular loss connect layers through processes. They do not require equal amounts or simultaneous changes. A measured amount is a stock at a time; a production or consumption rate is a change per interval under a stated account. Protein abundance alone also does not specify how much of that protein is active in the relevant setting.
The original two-state model supplies separate arbitrary measurement scales. RNA changes from four to eight, protein from eight to eight, and metabolite from six to three. Within-layer ratios are two, one and one half. Those ratios make the assigned changes comparable as fold changes while leaving the different underlying quantities distinct.
Fold changes compare each layer with itself
Read the diagram in words
RNA 8/4 gives two, protein 8/8 gives one, and metabolite 3/6 gives one half. The raw signals belong to different scales; their totals across layers have no assigned meaning.
| Molecular measurement layer | First assigned state on that layer’s scale | Second assigned state on that layer’s scale | Second signal divided by its own first signal |
|---|---|---|---|
| RNA signal | 4 | 8 | 8/4 = 2 |
| Protein signal | 8 | 8 | 8/8 = 1 |
| Metabolite signal | 6 | 3 | 3/6 = 0.5 |
| Second signal divided by its own first signal | 0 — 2 |
|---|
Raw signals have separate arbitrary scales
A fold change of one means unchanged signal
No flux, activity or causal direction is measured
Two original assigned states, different arbitrary raw scales for RNA, protein and metabolite signals. No actual molecules, rate, causal pathway, intervention, temporal duration or uncertainty model. Fold changes compare each layer only with its own first value.
Technical repetition is not a new biological unit
Technical replicates repeat some part of measurement using the same biological material. Biological replicates concern different biological samples under the intended design. The exact independent unit still depends on allocation, sampling and shared conditions. Merely assigning different file names does not create independence.
The original hierarchy has four stipulated independent biological units. Each supplies two repeated technical readings. The two readings can reveal measurement disagreement within that unit, but they cannot substitute for observing another independent unit. A comparison about a population needs an account of which population those units represent.
Here the paired readings average ten, twenty, thirty and forty. Averaging all eight readings happens to give the same mean as averaging the four unit means because each unit contributes equally. That equality does not make the eight rows independent or justify using an eight-unit uncertainty calculation.
Eight readings represent four independent units
Read the diagram in words
Readings 9/11, 19/21, 29/31 and 39/41 give unit means 10, 20, 30, 40. The four-unit mean and equally weighted eight-reading mean are both 25.
| Independent biological unit in this model | Technical observation within its unit | Assigned signal units |
|---|---|---|
| Biological unit one | 1 | 9 |
| Biological unit one | 2 | 11 |
| Biological unit two | 1 | 19 |
| Biological unit two | 2 | 21 |
| Biological unit three | 1 | 29 |
| Biological unit three | 2 | 31 |
| Biological unit four | 1 | 39 |
| Biological unit four | 2 | 41 |
| Independent biological unit in this model | Mean of this unit’s two readings |
|---|---|
| Biological unit one | 10 |
| Biological unit two | 20 |
| Biological unit three | 30 |
| Biological unit four | 40 |
| Independent biological unit in this model | 4 |
|---|---|
| Technical observation within its unit | 8 |
| Mean across four unit means | 25 |
| Mean across eight equally weighted readings | 200/8 = 25 |
Equal means do not establish eight independent units
Original four independent biological-unit account with two equally weighted technical observations per unit. Independence is stipulated, not inferred from files. No organism, intervention, test, uncertainty estimate or recommended replication number.
Combine evidence without erasing alternative explanations
Integration connects measurements through matched sample identities, definitions and context. A table joining RNA from one condition to protein from another is a different comparison from a table measuring both in corresponding samples. Tissue composition, time, technical processing and normalisation can all influence the recorded pattern.
A correlation or joint change can suggest a mechanism while leaving alternatives open. A common condition might affect several layers, a layer might change after a delay, or a measurement might be an imperfect proxy for the process of interest. Predicting a molecular amount and explaining why it changes are related tasks with different evidence requirements.
The two-state example supplies no intervention, rate measurements or uncertainty model. Its fold changes do not establish a pathway, adaptation or causal direction. They demonstrate why a measurement at one layer should not silently stand in for every other layer, and why a useful integration preserves rather than hides the missing evidence.
Worked example
Calculate changes at the correct level
Divide eight by four for the RNA fold change, eight by eight for protein, and three by six for metabolite. The results are two, one and one half. Each denominator is that layer's first-state value; none divides one type of quantity by a different layer's arbitrary scale.
For the four independent units, average nine/eleven, nineteen/twenty-one, twenty-nine/thirty-one and thirty-nine/forty-one. The means are ten, twenty, thirty and forty. Their total is one hundred, giving a unit-level mean of twenty-five.
The eight raw readings sum to two hundred and also average twenty-five. Equal weighting explains the agreement. The population-inference account remains four independent biological units, with repeated technical readings nested inside them; no test or sample-size recommendation follows.
Check your understanding
Does a twofold RNA change require a twofold protein change?
Show reasoning
No. They concern different molecular amounts and the processes connecting them. The toy record explicitly gives unchanged protein; it contains no evidence that would impose equal or immediate fold changes.
Do eight readings in the hierarchy provide eight independent biological replicates?
Show reasoning
No. Two readings belong to each of four independent units. Technical repetition informs measurement variation, while biological variation and population inference depend on the unit-level design.
Why does agreement between the two averaging methods not remove the nesting?
Show reasoning
Each unit contributes exactly two readings, so the means have equal weights. The arithmetic equality changes neither the shared origin of each pair nor the number of independent biological units.
What to carry forward
Integrate well-defined quantities and unit identities; an omic pattern can inform a mechanism without establishing it.
References
Resources supporting this lesson’s explanations and examples.