A biological signal can change dramatically without being the reason a phenotype changes.
Gene expression, protein abundance, phosphorylation, localization, and pathway-enrichment results can all provide useful evidence that a biological system has responded to an intervention. They may identify candidate pathways, prioritize experiments, or strengthen the plausibility of a proposed explanation. But observation of molecular change is not equivalent to demonstration of functional dependency.
This distinction becomes important when mechanistic language moves from ‘associated with’ or ‘modulated’ toward stronger claims such as ‘mediates,’ ‘drives,’ or ‘is required for’ an observed effect.
If a treatment increases the expression of a receptor, for example, the result does not establish that receptor signaling caused the physiological response. Increased abundance may accompany pathway activation, compensate for reduced activity, reflect downstream feedback, or simply occur in parallel with the true causal process.
Functional evidence asks a different question: what happens to the phenotype when the proposed component is specifically perturbed?
Inhibition, deletion, knockdown, mutation, pathway blockade, rescue, or carefully designed gain-of-function experiments can move an interpretation beyond molecular association. The strongest designs also test whether the proposed mechanism remains necessary across relevant biological contexts and whether plausible alternative pathways can explain the same observation.
Expression data therefore remain valuable, but their role should be calibrated correctly. They can support a mechanistic hypothesis. They do not, by themselves, convert that hypothesis into a demonstrated mechanism.
The scientific question is not only whether a molecule changed. It is whether the biological effect depends on that molecule changing in the way the proposed mechanism requires.