How to read this page

Entries are grouped by topic area. Each entry includes a brief description of why it's relevant to AIHS development. Links go to primary sources where possible (program homepages, consortium sites, foundational papers). The list is not exhaustive — it is the references actually drawn on in the study, plus a small set of additional starting points for further reading.

For depth on any specific advance, the individual advance pages list more specific ecosystem actors relevant to that capability.

Foundational genomics & epigenomics

  1. ENCODE Project Consortium — The Encyclopedia of DNA Elements; comprehensive functional annotation of the human genome. Foundational reference data for variant interpretation work in Bucket B. encodeproject.org
  2. Human Cell Atlas — International collaboration to build comprehensive reference maps of all human cells. Essential reference data for diagnostic interpretation. humancellatlas.org
  3. gnomAD (Genome Aggregation Database) — Aggregated and harmonised exome and genome sequencing data from large-scale sequencing projects. Reference catalogue for human variation underlying B1. gnomad.broadinstitute.org
  4. International Human Epigenome Consortium (IHEC) — Coordinated international effort to produce reference epigenomes. Underlies the data resources needed for B2. ihec-epigenomes.org
  5. NIH 4D Nucleome Program — Studying the spatial and temporal organisation of the genome inside the nucleus. Relevant to understanding regulatory mechanisms beyond linear sequence. commonfund.nih.gov/4dnucleome

Cellular and tissue mapping

  1. NIH HuBMAP (Human BioMolecular Atlas Program) — Building 3D atlases of human tissues at single-cell resolution. Core reference work feeding A1, A2, B2. commonfund.nih.gov/HuBMAP
  2. NIH SenNet (Cellular Senescence Network) — Mapping senescent cells across human tissues. Directly relevant to therapeutic targets in Bucket C. commonfund.nih.gov/senescence
  3. NIH SPARC (Stimulating Peripheral Activity to Relieve Conditions) — Bioelectronic medicine; maps of neural-organ interactions. Underlies A3 and parts of C4. commonfund.nih.gov/sparc
  4. NIH BRAIN Initiative — Comprehensive neuroscience programme including brain cell census, neural circuit mapping, and tools for in vivo neuroscience. Foundational for any AIHS neural capability. braininitiative.nih.gov

Predictive biology and AI

  1. AlphaFold (DeepMind / Isomorphic Labs) — Protein structure prediction at near-experimental accuracy. The proof of concept that foundation-model approaches can solve longstanding biology problems; cited throughout Bucket B. deepmind.google/technologies/alphafold
  2. AlphaMissense (DeepMind) — Classification of missense variants as likely benign or likely pathogenic. Direct precedent for B1. Cheng et al., Science (2023).
  3. ESM (Evolutionary Scale Modeling, Meta AI) — Protein language models for sequence-to-function prediction. Complementary to AlphaFold for B1-style work. github.com/facebookresearch/esm
  4. Perturb-seq — Combining CRISPR perturbation with single-cell RNA sequencing to map causal regulatory relationships. The leading current technique for the experimental side of B3. Dixit et al., Cell (2016); subsequent extensions.
  5. Joint Initiative for Causal Genomics (JICG) — Coordinated effort to scale perturbation biology for causal modelling. Direct relevance to B3.
  6. Physiome Project — International effort to develop mechanistic computational models of human physiology across organ systems. Underlies digital-twin work in B4. physiomeproject.org

Diagnostic technology

  1. Galleri test (GRAIL) — Multi-cancer early detection from a single blood sample using methylation patterns of cell-free DNA. Currently the most clinically advanced epigenomic liquid biopsy. Relevant to A5 and B2. grail.com
  2. Spatial Genomics Year of Methods — Nature Methods 2020. The methodological maturation that made high-resolution tissue molecular maps a research-grade tool, feeding into A1 and A2.
  3. Long-read sequencing (PacBio, Oxford Nanopore) — Technologies that resolve structural variants and complex genomic regions inaccessible to short-read methods. Foundational for comprehensive variant interpretation in B1.
  4. CardioMEMS (Abbott) — FDA-approved implantable heart-failure monitor. Real-world example of the chronic implantable sensor pattern targeted by A3.

Therapeutic delivery and gene editing

  1. Casgevy (Vertex / CRISPR Therapeutics) — First FDA-approved CRISPR-based therapy (December 2023), for sickle cell disease. Proof of concept for in-body gene editing.
  2. Lipid nanoparticle (LNP) delivery — The platform that enabled mRNA COVID-19 vaccines. Continues to be the most-developed delivery vehicle for nucleic-acid therapeutics; central to multiple C1 pathways.
  3. Voyager Therapeutics, Denali Therapeutics, Regenxbio — Companies developing AAV-based and antibody-shuttle technologies for crossing the blood-brain barrier. Direct relevance to C2.
  4. Wake Forest Institute for Regenerative Medicine — Pioneer of lab-grown organ tissues including engineered bladders implanted in patients. Foundational reference for C3. school.wakehealth.edu/WFIRM
  5. NervGen, BrainStorm Cell Therapeutics — Companies pursuing aspects of nerve regeneration. Representative of the small-but-growing C4 commercial ecosystem.

Synthetic biology and engineered cells

  1. Ginkgo Bioworks — Industrial-scale synthetic biology platform. Infrastructure relevant to engineered therapeutic agents in C1 and C5.
  2. Synlogic, Novome — Engineered probiotics with therapeutic functions. Demonstrate the bounded-agent design pattern central to C5.
  3. Senti Bio, Asher Bio — Programmable medicine startups developing logic-gated cell therapies. Direct precedent for multi-target coordinated intervention (C1).
  4. Voigt, Collins, Weiss laboratories (MIT) — Academic synthetic biology programs producing engineered cellular systems with sensor and response circuits. Foundational to A5 and C1.
  5. DARPA Safe Genes program — Programme to develop safety controls for engineered organisms, including reversibility mechanisms. Directly relevant to C5 and C6. darpa.mil/program/safe-genes

Federated learning and medical AI infrastructure

  1. MELLODDY (Machine Learning Ledger Orchestration for Drug Discovery) — Pharmaceutical-industry federated-learning consortium. Proof of concept that competing organisations can federate AI training on sensitive data.
  2. NVIDIA Clara — Federated-learning platform for medical AI. Infrastructure relevant to B5.
  3. OHDSI (Observational Health Data Sciences and Informatics) — Open-source informatics community with shared data models for federated medical research. Relevant to A4 and B5. ohdsi.org
  4. Owkin — Federated medical AI startup. Representative of the commercial ecosystem developing the practical infrastructure for B5.

Safety, governance, regulatory science

  1. FDA Software as a Medical Device (SaMD) framework — Current regulatory approach to AI/ML-based medical devices. Starting point for AIHS regulatory pathway development. fda.gov/SaMD
  2. Hastings Center — Bioethics research institute. Active programmes on emerging biomedical technology ethics relevant to the AIHS deployment landscape. thehastingscenter.org
  3. Nuffield Council on Bioethics (UK) — Comparable bioethics body with substantial output on genome editing, AI in medicine, and related topics. nuffieldbioethics.org

National research programmes

  1. NIH Common Fund — Funding mechanism for transformative cross-institutional programmes. Several CF programmes (HuBMAP, SenNet, SPARC, 4D Nucleome) are AIHS-relevant. commonfund.nih.gov
  2. UK Biobank, Our Future Health (UK) — Large-scale longitudinal population health studies. Reference data for population-scale variant interpretation in B1.
  3. NIH All of Us Research Program — US-equivalent large-scale longitudinal health study with diverse participant recruitment. Critical reference data for equity-aware AIHS development. allofus.nih.gov
  4. Canadian Institutes of Health Research (CIHR) — Primary federal health-research funding body in Canada, with relevant programmes on personalised medicine, regenerative medicine, and AI in health. cihr-irsc.gc.ca

Adjacent reading on the conceptual ladder

The protomolecule, as discussed throughout this study, is the fictional limiting case. Useful adjacent reading on the conceptual ladder from real biology to that limit:

  1. Prion biology — Prusiner's Nobel lecture (1997) and subsequent work establishing the protein-only hypothesis for prion replication. See the Institute's learning module M-Bio-01.
  2. Engineered self-replicators — Theoretical and experimental work on von Neumann self-replicating systems, from von Neumann's original theory through current synthetic biology realisations. The Institute's planned module M-Theory-01 addresses this.
  3. Directed panspermia — Crick & Orgel (1973), Icarus. The serious scientific proposal that earthly life was seeded. Useful as a conceptual orientation for what an engineered biological agent could in principle do given sufficient time and resources.
  4. Engineered Negligible Senescence (SENS / Aubrey de Grey) — A specific proposal for what aggressive intervention against ageing damage might look like. Speculative but technically engaged.