What I work on
I develop computational methods for reading, organizing, and comparing genomes — from models of gene regulation and RNA splicing to algorithms for annotation, assembly, and graph indexing.
Gene expression
I build sequence-to-function models that predict gene expression from DNA, and test whether what they learn transfers across genes, conditions, and species and points to real regulatory sequence.
RNA splicing
I build deep-learning models of splice sites and junctions, and use them as evidence for cleaner RNA-seq alignments, cross-species prediction, and variant interpretation.
Genome annotation
I develop LiftOn, which combines DNA and protein alignments to transfer accurate gene models onto new assemblies and related species, including complete and diploid human genomes.
Genome assembly
I assembled and annotated Han1, a gapless Southern Han Chinese genome, and contributed gene annotation to the complete diploid HG002 benchmark.
Pangenome indexing
I built tools to generate, recognize, and verify Wheeler graphs, the sequence graphs that admit BWT-style indexing, so that whether a pangenome graph can be indexed becomes something one can check.
