We developed LABEL (with Feng Bao’s group), a whole-mouse annotation model that combines a pre-trained pathology foundation model, UNI, with a spatially aware K-nearest neighbors (KNNs) classifier, enabling automated annotation of organs, tissues, and cell types on images from H&E-stained whole-mouse sections.

Original publication: Clevenger et al., Cell, 2026

We combined the two workhorses in gene expression from the beginning of the twenty-first century — microarrays and next-generation sequencing — to create Array-seq, a simple, large-format platform for spatial transcriptomics. Array-seq is (i) scalable thanks to its large format (11.31 cm²) and low cost, (ii) easy to adopt without special expertise or instrumentation, (iii) compatible with H&E staining—the gold standard for clinical pathology diagnosis—and (iv) readily applicable to all fields of basic and clinical research.

Original publication: Cipurko et al., Nature Methods, 2025.

We developed a method, 3-prime mRNA extension sequencing (PME-seq), for bulk RNA-seq profiling. We use a “fragment RNA first” approach followed by barcoded oligo(dT) priming for cDNA synthesis and sample pooling prior to sequencing library construction. While PME-seq is applicable to any RNA sample, including in low amounts (<1 ng per sample), we also created procedures to collect, store and lyse a dozen mouse organ types using conditions compatible with downstream high-throughput and low-cost RNA extraction and sequencing with PME-seq.

Original publications: Kadoki et al., Cell, 2017 & Pandey et al., Nature Protocols, 2020