The findings were showcased at the Keystone Symposia on Obesity Therapeutics, underscoring AI-augmented human genetics as a path to novel therapeutic targets.

Insitro unveils a first-of-its-kind AI-enabled, population-scale genetic analysis of brown adipose tissue (BAT) using an MRI-derived imaging phenotype and a BAT-focused GWAS in the UK Biobank cohort of nearly 70,000 participants.

Insitro frames this as enabling AI-driven human genetics in tissues difficult to study at scale, aiming to support differentiated obesity therapies that may avoid appetite suppression.

Using ClinML, the team derived a BAT imaging phenotype from UK Biobank MRI data by measuring the delta between abdominal and supraclavicular fat-signal fraction, revealing seasonal variation and links to cardiometabolic traits.

David Lloyd, Ph.D., emphasized a shift from trial-and-error discovery to AI-guided, phenotype-driven genetic discovery with potential impact on obesity therapeutics.

This approach overcomes PET limitations by deriving BAT-related phenotypes from widely available Dixon MRI data, enabling scalable human genetics for BAT.

Insitro presents itself as a causal AI company building the Virtual Human and TherML platforms, with substantial funding and partnerships guiding its neuroscience and metabolic disease therapy pipeline.

The work exemplifies a move from traditional drug discovery to scalable, AI-assisted human genetics and causal biology for therapeutic development.

A BAT polygenic risk score showed causal associations with multiple cardiometabolic traits, reinforcing BAT’s metabolic benefits.

Lloyd highlighted BAT’s growing role in metabolic health and ongoing exploration of peripheral fat reduction mechanisms during the presentation.

The study identified genetic loci linked to BAT biology and prioritized BAT-01 as a target; modulation reduced weight in diet-induced obese mice while preserving lean mass, indicating a peripheral mechanism distinct from central appetite pathways.

Insitro plans to continue evaluating BAT-linked genes from the GWAS using CellML and in vivo studies to broaden a pipeline of obesity and metabolic disease targets.