Benjamin Arya sits down with Omar Abudayyeh and Jonathan Gootenberg, the McGovern Institute duo behind foundational CRISPR-Cas13 systems and the SEND delivery platform, for a conversation that treats biology as a stack you can program. They open with a clear thesis: the field has moved past the question of whether we can edit genes and is now wrestling with how to engineer complex, multi-gene systems predictably. The centerpiece is the concept of the 'virtual cell' — a computational model trained on single-cell multi-omic data that simulates how a human cell responds to arbitrary perturbations. Instead of screening compounds or guide RNAs in a wet lab for years, researchers run thousands of in silico experiments, identify the most promising interventions, and validate only the top candidates. This collapses discovery timelines and changes the economics of biotech. The conversation then drills into the specific programmable systems they've built. They detail how Cas13's RNA-targeting mechanism enables transient, reversible edits — critical for aging interventions where you don't want permanent genomic scars. They explain SEND, which co-opts endogenous human retrotransposon proteins to package therapeutic RNA, sidestepping the immunogenicity that limits AAV-based delivery. On aging specifically, they decompose the problem into hallmarks — epigenetic noise, mitochondrial decline, senescent cell accumulation — and argue that each requires a distinct programmable modality applied in concert. The discussion is granular: they cover guide RNA design rules, delivery vehicle tropism, off-target minimization strategies, and the regulatory reality of polygenic therapies. Throughout, they return to the engineering mindset: specify the target state, model the intervention, measure the delta, iterate. There's no hand-waving about immortality — just a rigorous walk through the systems being built today and the decisions that determine whether they reach patients.
Key Insights
- Virtual cells trained on multi-omic data will soon let researchers run perturbation experiments in silico before touching a wet lab, collapsing years of trial-and-error into hours of compute.
- Programmable biology is shifting from single-target gene editing toward 'polygenic engineering' — tuning dozens of genes simultaneously to reverse complex aging phenotypes rather than treating one mutation at a time.
- The SEND (Selective Endogenous eNcapsidation for cellular Delivery) system repurposes human retrotransposon proteins to package and deliver RNA cargo, solving the immunogenicity problem that has plagued viral vector approaches.
- Aging can be decomposed into discrete, measurable hallmarks — epigenetic drift, mitochondrial dysfunction, senescent cell burden — and each is becoming druggable through orthogonal programmable systems deployed in combination.
- CRISPR-Cas13 targets RNA rather than DNA, enabling transient, dose-titratable interventions that avoid permanent germline changes — a critical safety feature for longevity applications where you want reversible control.
- The bottleneck in engineered longevity is no longer editing tools but delivery and tissue-specific targeting; lipid nanoparticles and engineered virus-like particles are the near-term platforms that determine what reaches the clinic.
Who should listen: Biotech founders and computational biology engineers building in silico discovery platforms or designing multi-target programmable therapies for complex disease.
Why This Matters
This conversation marks the moment when genome engineering transitions from a tools-building phase into a systems-integration phase — the same pattern we track across frontier compute, where raw capability without orchestration architecture fails to scale. The virtual cell concept Abudayyeh and Gootenberg describe is functionally a digital twin for biological systems, and the engineering decisions around delivery, reversibility, and multiplexing will determine which longevity startups actually ship.
