Stanford Model Discovers Appetite-Suppressing Peptide Independent of GLP-1

Researchers at Stanford University have deployed machine learning to identify an endogenous peptide that suppresses appetite via neural pathways independent of GLP-1 therapies.

Current anti-obesity pharmaceuticals such as semaglutide mimic glucagon-like peptide-1 (GLP-1). These drugs act on receptors in the brainstem and hypothalamus to delay gastric emptying and induce satiety. While clinically effective, GLP-1 receptor agonists frequently cause gastrointestinal side effects and treatment discontinuation. A notable proportion of patients also experience weight regain after stopping the regimen.

To locate alternative metabolic regulators, the Stanford team built a computational model to screen uncharacterised signalling peptides across the human proteome. The system evaluated amino acid sequences, predicted structural conformations, and scored binding affinities against neuroendocrine receptor targets. By filtering for molecules that bypass canonical GLP-1 pathways, the algorithm isolated a previously unannotated appetite-suppressing peptide candidate.

Preclinical trials confirmed that the candidate peptide reduces caloric intake and drives weight loss in animal models. Functional mapping showed that the molecule engages a separate subpopulation of neurons in the central nervous system. Crucially, the compound did not activate the specific circuitry linked to the nausea and gastrointestinal distress often observed with GLP-1 agonists.

The findings highlight how deep learning models can accelerate endocrine discovery. Traditional wet-lab methods struggle to isolate low-abundance signalling peptides from complex biological tissues, typically requiring iterative mass spectrometry and bioassays over several years. Machine learning models instead assess millions of sequence permutations in silico, identifying functional candidates prior to chemical synthesis.

The Stanford group is now modifying the peptide to extend its circulating half-life and improve oral bioavailability. Subsequent work will explore whether pairing this novel mechanism with reduced doses of existing GLP-1 drugs could yield additive metabolic benefits.

Will this computationally derived target maintain its efficacy and tolerability profile once evaluated in human clinical trials?