Building better ways to perturb, measure, and explain cell behavior.
My research focuses on the experimental and computational systems needed to identify causal biology—from scalable perturbation to dynamic measurement and mechanistic validation.
Virus–host biology provides a rigorous test environment for this work. Infection is dynamic, heterogeneous, and experimentally demanding: small differences in timing, cell state, or measurement strategy can change the conclusion.
I use this domain to develop platforms that compare perturbations across conditions, measure cell behavior over time, and translate high-dimensional observations into testable mechanisms.
Genetic screens become more useful when results can be compared across biological contexts rather than interpreted as isolated hit lists. I developed and analyzed a standardized CRISPRi platform interrogating more than 3,000 proteostasis-network genes across HSV-1, HCMV, and Vaccinia virus, progressing from survival-based assays to quantitative FACS readouts.
CRISPRiFACSComparative screens
PlatformA shared CRISPRi screening and analysis framework across three DNA viruses.
QuestionWhich host dependencies are shared, and which are specific to a given infection context?
OutcomeComparable gene-level evidence and prioritized candidates for mechanistic follow-up.
02
Dynamic single-cell measurement
Population averages obscure differences in when cells become infected, how strongly they respond, and which trajectories lead to different outcomes. V-SWITCH converts live RNA-virus infection into an OFF-to-ON fluorescent signal that can be followed over time in individual cells.
PlatformA single-vector reporter paired with Cellpose segmentation, Ultrack tracking, trajectory extraction, and response classification.
QuestionHow do infection kinetics and cell-state heterogeneity vary across individual cells?
OutcomeQuantitative live-cell trajectories validated with Dengue, Zika, West Nile, and HCoV-OC43, with public analysis code.
Live imagingSingle-cellReporters
V-SWITCH activation during live dengue virus infection. The elapsed-time counter shows hours post-infection.03
Computational and quantitative workflows
High-content experiments require infrastructure that preserves biological context while making large datasets practical to inspect and analyze. I build Python tools for CRISPR-screen exploration, microscopy processing, single-cell feature extraction, visualization, and machine-learning classification.
PlatformConnected workflows spanning data conversion, quality control, segmentation, tracking, visualization, and classification.
QuestionHow can experimental data be processed efficiently without hiding artifacts or silent failure modes?
OutcomeReusable internal workflows and a cell-death classifier trained on 224 live-cell trajectories, reaching 0.909 ± 0.015 ROC-AUC.
PythonMachine learningCellpose · Ultrack
04
Mechanistic and translational application
Platforms matter when they support stronger biological conclusions. I use orthogonal assays, quantitative imaging, molecular biology, mass spectrometry, compound screening, and relevant cell models to move from perturbation-screen evidence toward mechanism and therapeutic interpretation.
QuestionWhich host processes are both important to infection and experimentally actionable?
OutcomeMechanistic characterization of host dependencies and host-directed compounds across multiple viruses, including herpes- and poxviruses.
MechanismSmall moleculesVirus–host biology
Emerging direction
Toward more iterative experimental systems
An emerging direction in my work is to connect experimental design, laboratory automation, smart microscopy, and analysis more tightly. The goal is not automation for its own sake, but experiments that learn efficiently: measurements should reveal uncertainty, expose failure modes, and inform what to test next.
I am also prototyping AI-assisted workflows for hypothesis generation, criticism, evidence review, and scientific synthesis. I treat these systems as decision-support tools whose value depends on traceable evidence, explicit uncertainty, and validation through experiment.
