Aditya Pratapa, Ph.D.
About
Hello — I'm Aditya. I'm a Postdoctoral Associate at Duke University's Discovery AI initiative and the Department of Cell Biology, working with Rohit Singh and Purushothama Rao Tata.
Previously I was a Senior Data Scientist at Akoya Biosciences and at the Broad Institute of MIT and Harvard. My work sits at the intersection of computational biology, machine learning, and spatial biology — building tools that let otherwise incompatible measurements speak the same language.
Research interests
Selected projects
Transforming foundation-model representations for OOD data
Foundation models like scGPT promise to unify data across labs, yet their embeddings stay fragile under protocol shifts. USHER aligns embedding space via fused Gromov-Wasserstein optimal transport — no retraining — removing artifactual variation while preserving biology.
Read the preprint →Topology-flexible transforms for multimodal spatial omics
Serial sections deform and measure non-overlapping analytes. SAME aligns heterogeneous spatial data using histology and cell-type cues with controlled "space-tearing" transforms — unifying protein, RNA, and metabolite data.
Read the preprint →Benchmarking gene regulatory network inference
A benchmark that exposed limitations of popular unsupervised GRN methods — many barely beat random. Cited in nearly 800 works, it is now a community standard for fair GRN evaluation.
Read the paper → GitHub →Selected publications
Education