I'm a computational biology researcher building tools that read evolution like a manuscript, looking for the genes that scribble themselves into species together, and the ones that vanish in the same paragraph. My work focuses on Inherited Retinal Diseases: 40% of patients still receive no genetic diagnosis. The rest is a search problem.
For ≈40% of patients with inherited retinal disease, modern sequencing returns a verdict of unsolved. The mutations are there, somewhere; we are not yet asking the right question.
Most diagnostic pipelines read a gene the way you'd read a single sentence; does it spell a known disease? But genes don't function alone. They co-evolve in modules across hundreds of millions of years, leaving a graded fingerprint of conservation and loss across the tree of life. That fingerprint is where I look.
"Genes that travel together work together."
So I build the second look. Computational frameworks that compare ~2,000 eukaryotic genomes at once, normalize evolutionary signal against phylogenetic distance, integrate transcriptomics and HPO phenotypes, and rank candidates whose evolutionary biography matches the disease. The output isn't a final answer; it's a much shorter, much sharper list of suspects to take to the bench.
Build a continuous similarity profile for every human gene across ~2,000 eukaryotic genomes.
Detect genes whose normalized profiles co-vary across clades using NPP correlations.
Layer HPO ontology + transcriptomics to score candidates bidirectionally, gene↔phenotype.
Each row below is a gene. Each column is a species, sorted left-to-right by evolutionary distance. Cell intensity reflects a continuous conservation score, normalized against phylogenetic distance, and a vermillion column marks a clade where related genes' scores drop in concert; a co-evolutionary signature. Hover a row to read its profile. Hover a species to see how it participates across the matrix.
A continuous vector of normalized similarity scores; one per species; describing how conserved a gene is across the tree of life.
Two genes co-evolve when their normalized profiles rise and fall together across clades, suggesting shared function.
Human Phenotype Ontology, a controlled vocabulary describing patient symptoms, used to link genes to organs.
Five built & shipped projects. Click any row to expand the case study.
A pragmatic mix of computational and wet-lab craft. The dry side runs the analyses; the wet side keeps me honest about what biology actually does when you ask it a question.
Before the matrices, there were pipettes. Recombinant expression of bovine lactoferrin and human keratin in E. coli and Arabidopsis, in collaboration with Miruku at Prof. Oded Shoseyov's lab – including a co-authored chapter in Alternative Dairy Products and Technologies.
Designed and analyzed behavioral experiments on octopus learning strategies. Where my taste for data on stubborn, complex biological systems began.
↗ recommendation letterOpen to research collaborations, R&D roles in computational biology / machine learning, and conversations with curious people from any discipline.