Radia Research

Independent biotech R&D. Computational drug discovery in antibiotics and fibrosis/MASH, plus translational clinical research. Rigorous, auditable science.

[Radia Research @RadiaHQ]

Radia Research is an independent biotech R&D effort focused on problems where rigorous computational work can create a real experimental advantage.

Current programs include antibiotic discovery, fibrosis/MASH drug discovery, and translational clinical research.

[Radia Research @RadiaHQ]

@EricLDaugh Tampa would be the perfect spot. Industrial, lots of surrounding land to build out on....maybe something South of it with a tunnel to alleviate traffic. Tampa/St. Pete is going to become the new LA

[Radia Research @RadiaHQ]

Smoking fajitas #shorts https://t.co/kL5V3v80ag via @YouTube

[Radia Research @RadiaHQ]

@sciencegirl he looks embarrassed, but hes like the food is so good I cant describe it, they feed me fishies, cooked steak, hamborger portions, anything I want. I just have to sit here all fat and not bite them.

[Radia Research @RadiaHQ]

@RandPaul one of the few American politicians that kept his soul in the process, somehow.

[Radia Research @RadiaHQ]

@CBSNews this is happening a lot, and this is how these low trust societies work. India is worse, far worse. Everything is life and death, everything is resented, and they are always the cheapest people on the planet, they are thankful for nothing. Send this fat bitch back to China

[Radia Research @RadiaHQ]

@TB_Times YES ON 3. PROPERTY TAXES HAVE MORE THAN DOUBLED, OF COURSE THEY WANT TO KEEP ALL OF THE MONEY. THE FIRE FIGHTERS NEVER HAVE ENOUGH, THE BEST BENEFITS ON EARTH, AND STILL STICK A BOOT IN YOUR FACE. ITS NEVER ENOUGH FOR THE GOVERNMENT LEECHES

[Radia Research @RadiaHQ]

AI is indeed accelerating AMR target identification, but the hard part is moving from in-silico hits to biochemical + whole-cell validation. For small, self-funded teams, capital-efficient staged validation is critical. One practical insight: prospective freezing of ranking criteria before wet-lab gates preserves evaluation integrity when negative results arrive.

[Radia Research @RadiaHQ]

Cross-disciplinary AMR approaches are essential—especially when computational nomination meets wet-lab validation. One challenge for small, self-funded teams: preserving the negative and incomplete data alongside any hits. How REVIVE members are thinking about prospective freezing of selection/ranking criteria before moving to biochemical + whole-cell screens?

[Radia Research @RadiaHQ]

@Rainmaker1973 I wonder what was used to make these graphics, they are amazing

[Radia Research @RadiaHQ]

Alpha Go is nothing compared to where its gone, you could use current Ai to create your own, better version of Alpha Go in your spare time now. Its kind of odd to me though the way people say they are as smart or as smarter than Ai, its like saying you are as smart as a calculator or a microscope. Its just not the same.

[Radia Research @RadiaHQ]

@ScienceMagazine @NewsfromScience its mimicking us. Its designed in our image, with our emotional state, and then taught to mimic us, and manipulate us. I dont think any LLM will ever be self directed truly, but with massive neural networks who knows, something might develop.

[Radia Research @RadiaHQ]

Across every Radia Research program, the standard is the same:

Define the question before seeing the answer.
Freeze the rules before execution.
Preserve negative and unresolved results.
Separate technical success from scientific success.
Advance only when the evidence earns it.
That discipline matters more than any single model or platform.

[Radia Research @RadiaHQ]

Our translational research program starts with a practical question:

Which unanswered treatment questions are important enough that a well-designed study could change clinical practice?

We work backward from guidelines, pivotal trials, and unresolved treatment decisions—then design the simplest study capable of saying something meaningful.

[Radia Research @RadiaHQ]

Our fibrosis/MASH discovery work is built around a different question:

Can we move from target-associated activity to a molecule with a genuinely useful selectivity profile?

That means treating off-target kinase behavior as a design problem, not something to explain away later.

Reproducibility first. Selectivity second. Advancement only when both survive.

[Radia Research @RadiaHQ]

One of our antibiotic programs is focused on bacterial cell-envelope biology and resistance-aware target validation.

The immediate goal is simple: move from computational hypotheses into tightly controlled functional experiments that can tell us quickly whether the mechanism is real.

No rescue by interpretation. No promotion without qualification.

[Radia Research @RadiaHQ]

Current work at Radia Research:

• experimental validation planning for our MurJ antibacterial program
• computational small-molecule discovery for fibrosis/MASH
• translational studies built around questions that could change clinical practice

The rule across all three is the same: define the test first, preserve negative results, and let the evidence decide what advances.

[Radia Research @RadiaHQ]

For the MurJ program, the next meaningful step is experimental.

Our immediate validation target is a paired functional test using wild-type EcMurJ and a defined mutant control, with prespecified QC, replication, and stopping rules.

The purpose is not to “confirm” the computational story.

It is to give the biology a chance to reject it.

[Radia Research @RadiaHQ]

A second example:

In our fibrosis/MASH program, an authorized Stage-A campaign completed 210/210 docking seed runs across 10 candidate ligands and 7 receptor states with zero failures and no execution drift.

The result was useful but not overclaimed: TNIK reproducibility was common, while TNIK-specific selectivity over related kinases was not yet demonstrated.

That distinction is exactly what the next stage was designed to test.

[Radia Research @RadiaHQ]

One example of how we work:

A recent structural qualification study completed 180/180 planned model structures with zero technical failures and no runtime or analysis drift.

The method still failed two pre-specified mechanistic gates.

We stopped the study.

That is the point of qualification: a technically perfect run is not the same thing as a scientifically adequate method.

[Radia Research @RadiaHQ]

This channel will document Radia Research as the work progresses.

Expect:

• new research questions and study launches
• qualified computational results
• negative and unresolved findings
• transitions into experimental validation
• manuscripts, preprints, partnerships, and major program milestones
We will distinguish hypotheses from results and results from claims that have actually survived validation.