Zachary J. DeBruine

This paper identifies an EZH2-driven immune evasion phenotype in FUS::ERG AML that is reversible with tazemetostat, offering a potential targeted therapy for high-risk acute leukemia.

The paper presents a scalable Non-negative Matrix Factorization of the Human Cell Census to create interpretable transcriptional programs for biological reasoning.

The paper shows that preserving both spliced and unspliced RNA data in single-cell analysis reveals more latent structure than using either alone or their sum.

The paper shows that early life epigenetic variation can prime cancer-susceptibility states that persist into adulthood and correlate with poor prognosis in human cancers.

An Rcpp implementation brings non-negative matrix factorization to singular-value-decomposition-like runtimes on sparse single-cell matrices, adding a scaling diagonal that keeps regularization penalties consistent across random initializations.