Download Champions' SITC 2025 Poster #148
DRUG-seq in CTG3D Organoid Models: A High-Throughput, Transcriptomics-Driven Platform for AI-Enhanced Drug Discovery
Poster #148, presented at SITC 2025
Traditional high-throughput screens rely on viability and lack mechanistic resolution. DRUG-seq merges compound screening with bulk transcriptomics to deliver rich MoA signals at scale. We adapted the workflow from cancer cell lines to Champions’ CTG-3D patient-derived organoids, seeding 384-well plates, dosing with standard-of-care agents, and using UMAP and clustering to resolve reproducible transcriptional fingerprints. The assay detects over 13,000 genes and hundreds-of-thousands of UMIs from fewer than one thousand cells.then integrates AI and public connectivity resources to classify MoA and predict responses.The workflow uses in-well lysis and reverse transcription with barcoded oligo-dT primers, pools cDNA across wells, prepares Illumina libraries, and sequences on a NextSeq 2000 at 1 to 2 million reads per well. The platform supports more than 50,000 perturbations per week across small molecules and genetic tools. Analysis pipelines generate gene-by-barcode matrices, normalize and visualize with PCA or UMAP, and compare signatures to large public perturbation sets to confirm MoA and reveal similarities for repositioning.
Learning outcomes:
- Understand how DRUG-seq on Champions’ proprietary bank of patient-derived organoids produces tightly clustered transcriptional signatures that match known MoA classes and thusprovide relevant biology not captured in immortalized cell lines.
- See how the assay scales in 384-well plates, while enabling high sensitivity and reproducibility.
- Recognize that siRNA knockdowns generate expected downregulation of target genes, confirming functional specificity in organoid systems.
- Learn how AI/ML and connectivity mapping tools, for example iLINCS (Library of Integrated Network-Based Cellular Signatures), classify compound activity, support prioritization, and inform prediction of unscreened agents.
- Identify practical next steps, including exploratory screens in patient-derived organoids and the use of generative models to design compounds with desired transcriptomic profiles.
Download the Poster