Cardiomyocyte Multimodal Reference Dataset, built for cardiac safety teams.
CMRD: target ~80 compounds (including post-2024 FDA NMEs) × patient-derived iPSC-cardiomyocyte lines × paired Cell Painting and DRUG-seq × dose-response. Pre-launch partner program open for inaugural pharma collaborators.
what we offer
CMRD
Pre-launch partner program for pharma cardiac safety teams. Co-design the compound list. Early access during build (Q4 2026 to Q1 2028). 6-month exclusivity window before public release. NDA-protected.
calibration program
Run your proprietary compounds through the same iPSC-CM pipeline that builds the CMRD. Multimodal mechanism profile + comparison to CMRD reference. NDA-protected. Available alongside or independent of dataset partnership.
academic partnerships
CRISPR or small-molecule perturbation screens, end-to-end wet-lab + analysis. Grant-funded co-development with PIs whose labs don't have the bandwidth.
The published iPSC-CM cardiotoxicity reference data covers fewer than 40 compounds across the four largest public datasets (Hansen 2024, Sharma 2017, Burridge 2016, Wu 2024), is single-modality, lacks post-2024 NMEs, and has limited patient-line dimension. CMRD fills that gap as a defined, partner-funded reference dataset. The work extends an established cardiac biology direction (cardiotoxic compounds suppress cardiac-identity gene transcription, established by Ito 1990 and confirmed at chromatin layer by Matthews 2025) into a multi-class, magnitude-controlled, operationalized screening biomarker. Methods paper preprint forthcoming.
CMRD specs
what's in the dataset
~80 compounds (target): textbook cardiotoxic + textbook non-cardiotoxic + ~20 post-2024 FDA NMEs (curation methodology; specific sourcing in active scoping) + cardiac-targeted therapeutics + partner-nominated compounds.
NDA + scope. Compound, target endpoints, and the question handed off.
->
Iterative rounds
2–3 cycles
Each round of data informs the next round of conditions tested.
->
Week 4–8
Delivery
Mechanism profile on your compound, calibrated against the CMRD reference where applicable.
inside each cycle
01 design
AI selects
Next round's experimental design: perturbations, concentrations, conditions.
02 execute
Cloud lab
Plate runs end-to-end on cloud-lab automation. No manual handoffs.
03 measure
Multimodal
Cell Painting and DRUG-seq, co-measured (morphology + transcriptomics).
04 analyze
Model updates
Each round's results inform the next round's design.
↻ Output of cycle N is input to cycle N+1.
Each round informs the next. Round 1 results select Round 2 conditions (dose, timepoint, modality emphasis) so that the engagement converges on the highest-value answer for the cardiac safety question you came in with.
founded by
Daniel Reda. Two prior exits in life science data: CureTogether (acquired by 23andMe) and Redasoft (acquired by Hitachi). Background in Molecular Genetics.