Platforms & Projects
A selection of experimental platforms, public analysis pipelines, internal tools, and ongoing development. Each project is labeled by maturity so completed work and active projects remain distinct.
Flagship platform
A cell-encoded, single-vector OFF-to-ON fluorescent reporter that makes live RNA-virus infection measurable at single-cell resolution.
Fluorescently tagging RNA viruses can reduce viral fitness and produce unstable constructs, while existing cell-encoded reporters often depend on artificial protease overexpression.
The system needed low background, infection-dependent activation, a single-vector implementation, and compatibility with live imaging without engineering the virus.
V-SWITCH uses infection-dependent fluorescent-protein reconstitution and a pipeline for segmentation, tracking, normalization, trajectory fitting, and response classification.
Validated with Dengue, Zika, West Nile, and HCoV-OC43. Fluorescence intensity tracks viral RNA levels and resolves activation timing across individual cells.
Supports live infection measurement, antiviral screening, functional genomics, and analysis of heterogeneous outcomes without a fluorescently engineered virus.
Experimental reporter and Python workflow for detecting and quantifying live RNA-virus infection. The pipeline integrates Cellpose segmentation, Ultrack tracking, fluorescence-trajectory extraction, sigmoid fitting, activation detection, and response classification on multidimensional microscopy data.
Standardized CRISPRi screening across more than 3,000 proteostasis-network genes in HSV-1, HCMV, and Vaccinia virus. An internal Python dashboard supports hit ranking, gene-level statistics, quality review, and cross-virus comparison.
A developing machine-learning workflow for classifying cell-death outcomes from time-resolved single-cell measurements. The work emphasizes trajectory features, interpretable quality control, and distinguishing biological transitions from imaging or tracking artifacts.
An actively developed 13-stage workflow in which specialized agents propose hypotheses, challenge assumptions, review literature-grounded evidence, and synthesize conclusions. Deterministic guardrails reduce fabrication risk, while a 142-run audit tracks where the system’s reasoning and validation checks still fail.