Shift Bioscience, a Cambridge, UK, biotechnology company uncovering the biology of cell rejuvenation to develop new therapies for age-driven diseases, has announced the publication of new research in Nature Biotechnology [1]. The peer-reviewed study describes an improved calibration framework for deep learning-based genetic perturbation models, a type of AI virtual cell.

Questions over model reliability

Genetic perturbation models are designed to predict how cells respond at a transcriptomic level to genetic interventions, including the activation and inhibition of genes. They can support scalable in silico target screening, but previous studies have questioned their reliability, with some models failing to outperform simple baseline approaches.

The new study builds on foundational research reported by the company in November 2025.

It defines a reliable benchmarking system that considers the biological and technical signals in a dataset, providing more meaningful insight into model performance. The framework demonstrated that underperformance in some past benchmarks could stem from miscalibration of the metrics used to compare models, which reduced sensitivity to genuine model performance.

Screens to focus on fibrosis

Shift will now use the findings to launch large-scale in vitro and in silico screens for novel, dual-purpose inhibition targets, with potential application in rejuvenation and the treatment of age-related disease. Following the discovery of SB-101, the company’s first dual-purpose target, the screens will focus initially on uncovering targets for both rejuvenation and the treatment of fibrosis, which Shift describes as a key driver of ageing and age-related disease.

Route towards clinical development

Dr Brendan Swain, chief scientific officer and founder of Shift Bioscience, said: “Our findings show that by using well-calibrated metrics and the right dataset, virtual cell models can generate biologically meaningful insights. As a result, we can use them with greater confidence to identify promising new targets that are relevant to ageing and disease. We are applying this framework directly in our target identification programme, focusing on targets whose inhibition can support both rejuvenation and treatment of age-related disease, giving us a clearly defined route towards clinical development.”

For more information, visit: https://shiftbioscience.com/

Reference:
1. Miller, H.E., Mejia, G.M., Leblanc, F.J.A. et al.
Deep learning perturbation models can outperform baselines on calibrated metrics.
Nat Biotechnol (2026).
https://shorturl.at/fT4Lu