Baylor's GluePlex predicted where a molecular glue grips VAV1 before the lab confirmed it
A Baylor College of Medicine team used a structure-prediction workflow called GluePlex to find where a molecular glue grips the protein VAV1 before any experiment showed it. A separate group later reached the same conclusion in the lab.
The Scientist · Science desk
What happened
- Follow-up experiments showed the compounds act through the cell's protein-recycling system and depend on cereblon, a hub of the degradation pathway.
- The GluePlex workflow modeled how VAV1, cereblon and the glue assemble into a single three-part complex.
- The model named VAV1's SH3-2 domain as essential, and lab tests narrowed the glue's grip to a small surface loop that acts as a degron.
Why it matters
- What's now possible capability Structure prediction becomes useful at the optimization stage, before a complex is solved, so chemists can act on a binding mechanism the bench has not yet confirmed.
- What it sets up precedent A cereblon glue that reads a non-G-loop degron makes more proteins plausible degradation targets than the field had assumed.
- What it limits constraint The evidence is degradation in cells and one confirmed contact point, so whether removing VAV1 is safe and effective in a body remains open.
Next story · Product · Sep 7
Six independently built ageing clocks agree on the same 42 patients from Insilico's lung trialConfirmed3 publishers
The order of operations matters here. GluePlex, a workflow built from AI, protein-structure prediction and physics-based modeling [14], reached the glue's contact point on VAV1 without a solved structure of the assembled complex, and the lab confirmed the prediction afterward [19]. A separate group, working from experiments, has since reported the same conclusion about VAV1 [18].
The second result concerns the shape of that loop. Most cereblon-dependent glues were thought to need a structural feature called a G-loop, and VAV1's degron is a different structure [17]. The authors read that as widening the set of proteins cereblon glues might one day degrade [20].
VAV1 is worth removing because of where it sits. It is found mainly in immune cells, where it helps activate T cells, and abnormal activity feeds T-cell lymphomas and chronic inflammatory disorders [8]. Traditional drugs block one function of a protein. Degraders remove the whole protein, and the team argues that can give a more complete effect [9]. "These compounds bring a target protein to the cell's natural protein-disposal machinery, which destroys the target," said Wang, who directs Baylor's Center for NextGen Therapeutics [6][5].
The compounds came from a screen. The team used high-throughput proteomics, which reads thousands of proteins at once [11]. Lin, a postdoctoral researcher in Wang's lab, said the analysis "revealed a series of compounds, including NGT-201-12, that caused VAV1 levels to drop while affecting relatively few other proteins" [10][12]. Follow-up work tied that degradation to cereblon [13].
The study shows degradation in cells and one confirmed contact point in a model, published in Nature Communications [2][16]. It does not show that removing VAV1 treats a lymphoma or an autoimmune disorder in an animal. And it confirms one prediction on one complex, not a success rate across many targets. The paper's broader claim, that this modeling lets chemists refine compounds before experiments reveal how they act, rests for now on this one worked example [4].
What to watch
- Whether GluePlex predictions hold up when tested against many target proteins, not just VAV1, which would establish an actual hit rate.
- Animal studies testing whether degrading VAV1 changes disease in T-cell lymphoma or inflammatory models.
- How the medicinal-chemistry work on NGT-201-12 changes the compounds' potency and selectivity.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence72
- Adoption15
- Hype gap+8
- Incentives68
- Confidence62
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
A team led by Baylor College of Medicine developed a strategy that combines analysis of thousands of proteins with artificial intelligence to speed the discovery of small molecules called molecular glues.
- [2]
The work was published in Nature Communications.
- [3]
The approach uncovered a new class of molecular glues that could neutralize harmful proteins linked to blood cancers and autoimmune diseases.
- [4]
The study reports that AI-based structural modeling can help chemists optimize compounds well before experiments reveal how they work.
- [5]
Senior and co-corresponding author Dr. Jin Wang directs Baylor's Center for NextGen Therapeutics and is an endowed professor in pharmacology and in the Verna and Marrs McLean Department of Biochemistry and Molecular Pharmacology at Baylor.
- [6]
These compounds bring a target protein to the cell's natural protein-disposal machinery, which destroys the target.
- [7]
The team's molecular glues selectively remove a protein called VAV1.
- [8]
VAV1 is found mainly in immune cells, where it helps transmit signals that activate T cells, and abnormal VAV1 activity can contribute to diseases such as T-cell lymphomas and chronic inflammatory disorders.
- [9]
Traditional drugs typically inhibit only one function of a protein, while targeted protein disposal eliminates the entire protein, potentially providing a more complete therapeutic effect.
- [10]
First and co-corresponding author Dr. Hanfeng Lin is a postdoctoral researcher in the Wang lab.
- [11]
The team screened a library of molecules using high-throughput proteomics, a technology that can assess thousands of proteins simultaneously.
- [12]
This unbiased analysis revealed a series of compounds, including NGT-201-12, that caused VAV1 levels to drop while affecting relatively few other proteins.
- [13]
Follow-up experiments confirmed the compounds worked through the cell's natural protein-recycling system and relied on the protein cereblon (CRBN), a key component of the protein-degradation pathway.
- [14]
The researchers combined artificial intelligence, protein-structure prediction tools and physics-based modeling into a computational workflow called GluePlex.
- [15]
GluePlex predicted how VAV1, CRBN and the molecular glue come together to form a three-part complex.
- [16]
The model identified VAV1's SH3-2 domain as essential for degradation, and experimental tests confirmed the prediction and pinpointed a small surface loop on VAV1 that acts as a degradation signal, or degron.
- [17]
The surface loop on VAV1 is different from degradation signals commonly associated with cereblon-targeting molecular glues.
- [18]
A separate research team has since reported a similar conclusion about VAV1 using different experiment-based methods.
- [19]
The computational workflow arrived at the contact point on its own; the prediction was made without reference to any experimental structure of the complex and was subsequently confirmed in the lab.
- [20]
Most known cereblon-dependent molecular glues were thought to require a structural feature known as a G-loop; together with the independent report, the study indicates cereblon can also recognize a different structure, broadening the range of proteins that may be targeted for degradation.
Sources
2 independent publishers whose own reporting we read for this story.
- genengnews.comAI and Proteomics Accelerate VAV1 Molecular Glue Discovery
1 article · September 29, 2026
- phys.orgAI structure prediction speeds discovery of 'molecular glues' to treat disease
1 article · September 29, 2026
Topics and entities
Entities
- Center for NextGen Therapeutics
- Nature Communications
- Baylor College of Medicine