Getting there is a physical problem.
A compound has to cross tissue before it can act on anything. That crossing is governed by physics that a well plate removes, and it is the first place a promising molecule quietly stops working.
The problem
In a well, every cell is the same distance from the drug: zero. The compound is in the medium, the cells are in the medium, and the only variable left is chemistry. That is what makes the assay reproducible, and it is also what makes it silent about tissue.
A tumour is not a well. A molecule leaves a vessel, crosses the vessel wall, and then has to move through packed cells and matrix to reach anything further in. Meanwhile the tumour is pushing back: raised interstitial fluid pressure drives flow outward, against the direction the drug needs to go. And the compound is being cleared the whole time.
What LURA computes
The Spatial Twin solves the transport problem on a three-dimensional geometry rather than reasoning about it by analogy. The compound is released from the vascular boundary and moves under diffusion, against convection driven by the pressure field, while clearance removes it.
- Diffusion through the interstitial space, set by the compound and by the density of the tissue it is crossing
- Convection from the interstitial pressure field, which in a tumour usually opposes penetration
- Clearance and binding, which decide how much of what enters is still free to act
- Time, because a concentration that arrives after the window has closed is not an exposure
The output is not a single number. It is a concentration field: what the drug level is, at every position in the modelled tissue, over the course of the schedule.
What comes out
From that field the questions a project actually argues about become measurable rather than assumed. How far in does this compound get. Does a different dose or a different schedule change that, or only change the total given. Two candidates with similar potency — does one of them reach further than the other, and by how much.
Because the field is spatial, a shortfall has a location. That matters more than it sounds: a compound failing everywhere and a compound failing only in the poorly perfused core are two different problems with two different fixes.
What this does not claim
Transport is solved as physics on a modelled geometry, not measured in a patient. The quality of the answer depends on the quality of the tissue description it is given, and LURA reports which coefficients came from measurement and which came from a class prior.
It is a model of getting there. Whether a compound is selective once it arrives is a different question, answered by a different engine, and the two are never merged into one score.