A macrocycle can fit before it can travel

A designed macrocycle can produce a satisfying image: the predicted shape and the experimental structure appear to agree. For medicinal chemistry reporting, I would pause at that moment and label the achievement precisely. What did the structure confirm? Which proposed properties still need a different measurement? The picture should start that conversation rather than finish it.

A binding design reaches experiment

Rettie and colleagues introduced RFpeptides in a 2025 study. They tested at most twenty designed macrocycles against each of four proteins and obtained binders for every target. Crystal structures for three complexes closely matched the design models. In the discussion, simultaneous optimization of binding and cell permeability or oral bioavailability is described as a future possibility.¹

The authors also note that computational rankings did not perfectly reproduce the ranking of experimental affinities.¹ That distinction matters when selecting an optimization starting point: structural agreement and rank-order accuracy are different claims. This article examines those evidence boundaries rather than reviewing the network architecture.

Decide what a change is meant to preserve

For a follow-up campaign, I would ask the team to state which feature of a successful design it wants to keep. The answer could be a particular binding geometry, a measured affinity or a practical feature of compound preparation. I would then ask which new property the next molecular change is intended to address.

Consider a hypothetical analogue chosen to investigate membrane passage. I would want its predicted advantage written down before measurement, along with the binding result that the team hopes to retain. If the analogue loses the desired binding behavior, the report should say so directly. Calling the change an improvement without naming the endpoint would conceal the central trade-off.

This is a proposed reporting approach, not evidence that the published binders have undergone that campaign. It is also why I would avoid presenting a bound-state picture as a general certificate of developability. A strong result deserves a strong, narrowly defined claim.

Keep models and measurements side by side

My preferred article would place each proposed interaction beside the observation that tests it. If a model ranks one design first and the assay favors another, that disagreement should be preserved. I would ask whether it changed the next selection and whether the team could explain the difference without fitting a new story to every outcome.

The same record should identify which molecules reached measurement, which did not and why. A failed preparation should not silently disappear into the category of an uninteresting design. Those distinctions would help readers see what the design system contributed and where experimental judgment remained necessary.

RFpeptides makes the binding-and-structure part of this conversation concrete.¹ Our editorial conclusion is a request for the next piece of evidence: follow one design through a deliberate property trade-off while retaining the confirmed result that made it worth optimizing. A macrocycle's fit and its ability to reach a biological site should remain separately stated questions until the relevant experiments connect them. That boundary gives a successful design a useful starting point for medicinal chemistry, without promising a finished medicine.