Stable-Shift: Biologically Structured Prediction of Transcriptional Responses to Unseen Gene Perturbations

Name: Stable-Shift: Biologically Structured Prediction of Transcriptional Responses to Unseen Gene Perturbations
Uploaded: 2026-09-16T09:07:40.750250+00:00
Duration: 414 s
Description: Researchers Sajib Acharjee Dip and Liqing Zhang developed Stable-Shift to predict how cells respond to unseen gene perturbations. By integrating biological maps like STRING protein interactions and Gene Ontology via graph convolution, the model outperforms existing models like GEARS on the K562 Perturb-seq benchmark, improving accuracy in predicting transcriptional responses.

Researchers Sajib Acharjee Dip and Liqing Zhang developed Stable-Shift to predict how cells respond to unseen gene perturbations. By integrating biological maps like STRING protein interactions and Gene Ontology via graph convolution, the model outperforms existing models like GEARS on the K562 Perturb-seq benchmark, improving accuracy in predicting transcriptional responses.

Open the full episode