Transformer model predicts cell responses to psilocybin drug effects
A Transformer-Based Delta Expression Encoder for Psilocybin Transcriptional Response: Architecture, Representations, and Biological Validation
Machine Learning
Summary
People react differently to the psychedelic drug psilocybin, and this depends on how it changes gene activity in brain cells. The authors trained a type of AI called a Transformer on detailed gene data from brain cells to predict if genes are turned on or off after taking psilocybin. Their model can guess these changes reasonably well without needing pre-existing knowledge about biological pathways. They found that gene activity patterns vary a lot by cell type and that some expected biological links did not hold true, offering new insights into how psilocybin affects the brain.
What this means in practice
- •For pharmaceutical developers: Develop cell-type specific biomarkers based on gene activity patterns to evaluate psilocybin’s effects during drug development.
- •For bioinformatics software engineers: Integrate Transformer-based models for interpreting single-cell gene expression changes in complex drug response datasets.
Authors
Sai Jayakumar
Abstract
Understanding why individuals respond differently to psilocybin requires modeling the drug's transcriptional perturbation signature at the cell-type level. I present a Transformer-based delta expression encoder that learns to classify differential gene expression status - upregulated, downregulated, or neutral - from single-nucleus RNA-sequencing data, without supervision from pathway annotations or prior biological knowledge. The model is trained on pseudobulk profiles from 623 examples spanning 18 cell types, 2 drug conditions, and 6 timepoints derived from the Liao et al. 2025 dataset, and achieves 69.4% weighted classification accuracy. Three principal findings are reported, alongside one direct test of a published hypothesis that returned a result inconsistent with that hypothesis. First, per-cell-type classification accuracy ranges from 28.3% (L2/3 IT, a primary HTR2A-expressing psilocybin target) to 99.6% (endothelial cells), consistent with known psilocybin response biology. Second, psilocybin-induced transcriptional downregulation is significantly more stereotyped across individuals than upregulation (Mann-Whitney U=18615.0, p<0.0001), a novel finding with a cortical depth gradient across excitatory subtypes. Third, attention-guided gene co-regulation analysis recovers drug-specific modules without pathway supervision. Separately, a direct test of whether baseline HTR2A expression predicts drug-response separability across cell types found a significant negative correlation (Spearman r = -0.7088, p = 0.0021), the opposite of what a simple HTR2A-gating account would predict.