Joshua Yang

Biology has a data imbalance: genomic and transcriptomic data are abundant, while protein data remain sparse—even though proteins are the functional layer of biology. I encountered this directly through my work on organ injury and transplant rejection. Mass spectrometry revealed important biology, but its data were far harder to generate and process than DNA or RNA sequencing. That experience convinced me that biological AI needs a much richer protein data layer. My work in protein sequencing aims to close that gap by measuring protein identity at the scale required to build datasets, discover patterns, and train better models.

Organ injury was often detected too late and with tools poorly suited to continuous monitoring. Many cell-free DNA approaches relied on expensive sequencing of blood to distinguish donor-derived from recipient-derived fragments. We took a different view: urine comes directly from the kidney and serves as a molecular window into the transplanted organ. In this setting, urinary cfDNA came from the transplant, allowing inexpensive measurements such as total quantification to become informative without resolving two genomes. That insight led us to combine cfDNA, urinary biomarkers, gene expression, metabolomics, and immune biology—moving from measuring rejection toward predicting and ultimately controlling it.

My curiosity has repeatedly taken me beyond the boundaries of a single field—from kinase signaling and innate immunity to unusual clinical cases, biomedical instrumentation, and technological foresight. These projects may appear unrelated, but they share a method: notice what does not fit the prevailing explanation, pursue the underlying mechanism, and build whatever tools the question requires. Each taught me ways of thinking that transferred to the next. I value this kind of curiosity because foundational or seemingly peripheral questions often become unexpectedly important when viewed from a different problem.