GeneSpeak-FP: Target and Compound Retrieval from Observed Cell-Level Perturbation Signatures
2026-07-20 • Machine Learning
Machine Learning
AI summaryⓘ
The authors developed a model called model to figure out which drug and target might have caused a specific change in a cell's gene activity. They trained their model using data where the response of cells treated with drugs was compared to untreated cells. Their model could successfully identify the correct drug targets and compounds within a fixed set, but it is not yet proven to work on totally new drugs or cell types. This work helps match observed cellular responses to known treatments in large datasets.
single-cell perturbationTransformer modeltarget retrievaldrug screeninggene expression signaturecell-line referenceRecall@10compound identificationmulti-task learningclosed-library setting
Authors
Kseniia Vaniushkina, Jeongmin Lim, Jinyong Park
Abstract
Large-scale single-cell perturbation atlases make it possible to ask an inverse question: given an observed transcriptional response, which annotated targets and compounds in a fixed library are most consistent with that response? We present \model, a Transformer retrieval model for this closed-library setting. Each input is a cell-level perturbation signature formed by contrasting one treated cell with a cell-line-specific mean DMSO reference. The encoder maps the signature to a target-retrieval vector and a molecular-embedding vector, trained jointly with supervised target losses and structure--transcriptome alignment. We evaluate on Tahoe-100M conditions with mapped target annotations using a within-compound stratified 90/10 condition-pair split of 10,505 training and 1,168 validation drug--cell-line pairs. Because compounds and cell lines can occur in both partitions, the experiment measures held-out condition-pair retrieval rather than generalization to unseen compounds or cellular contexts. In a Monte Carlo evaluation over 38,400 sampled validation cells, \model\ achieved target Recall@10 of 0.408 and Recall@20 of 0.544, together with compound Hit@1 of 0.129, Hit@10 of 0.343, and mean reciprocal rank of 0.205 over a 379-compound bank. A separate diagnostic evaluation produced nearly identical values for the main model and large gains over a random-vector control and post-hoc bag-of-genes controls. These results demonstrate that a single multi-task model can recover both mapped target annotations and recorded compound identities from observed cell-level responses in the evaluated Tahoe-100M closed-library setting. Generalization to unseen compounds and cellular contexts remains to be established.