Bring your compounds. LURA shows which are selective in which tumor contexts, which reach the cells that matter, where tissue breaks the story, and what to advance and what to stop, before years and millions disappear into the wrong program.
Bring us the decision you can't makeFor pharma, mid-sized biotech and CROs.
A portfolio in. A field compiled. A picture you can defend.
Around nine in ten candidates that enter clinical development never reach approval. Bringing one drug to market costs billions of dollars and well over a decade. Almost none of that is the price of learning that something works. It is the price of learning, years late, that something does not.
The evidence that would have said so almost always existed earlier. It was never computed.
A potent compound that never arrives is a failed program with good numbers. A well plate cannot see that. Most programs find out in the animal, years and millions of dollars late.
Drug–cell pairs screened. 2.5 million drug–cell interactions evaluated in 1.5 seconds on a single NVIDIA A100.
You bring the portfolio. LURA reads it against the tumor, and answers in seconds. Every program you stop early gives back its months and its millions while they still matter.
Not one molecule at a time.
Not one cell line, one afternoon.
On the same page, each labelled for what it is.
Which compounds win in which genotype, and whether they reach the cells that matter.
Whether the payload reaches the cells the antibody never touched.
Where the evidence disagrees, you see it. It is not averaged away.
With the reasons attached.
Both auditable by your own scientists.
A compound cannot work where it never arrives.
From a solved tissue model, not a measurement. Every result says which.
In a well, every cell sits at zero distance from the drug. In a tumor it has to travel. The outer layers get treated. The core never sees a dose.
Oxygen, acidity and pressure change across a mass, and each changes what a drug does. One compound, two answers.
The share of the tumor at a therapeutic dose, and the regions that never get one, by name. A named failure is a design brief. An averaged one is a surprise in year three.
You know before the animal study.
Compounds ranked by selectivity in each genetic context.
A labelled estimate with a range. Never dropped, never dressed up as a measurement.
The direction of the answer, with its sensitivity printed beside it.
Every number with its evidence, its assumptions and its limits.
A sealed record you can reopen.
In a retrospective test, EGFR inhibitors ranked hundreds of times more selective than controls in an EGFR-mutant line. BRAF inhibitors did the same in a BRAF-mutant line. A line with no targetable driver showed no signal.
A computational answer nobody can challenge is not evidence. It is an assertion with arithmetic attached. So every LURA result carries the evidence behind it, the assumptions under it, its uncertainty, and what is still unknown. A direct measurement never reads like an inferred one.
A series that is not separating. A genotype you cannot explain. A candidate that works in a well and fails in tissue.
The first mission is a program you have already closed. We run it. We show you where LURA agrees with your bench result, where it does not, and how many months and millions it could have taken out of the road.
If a study is live and the clock is running, say so in the last field. Same-day answer.
The people behind it.
About Elyris Labs