Embryo Project
Beyond today’s polygenic ceiling
Embryo Project aims to push genotype-to-phenotype prediction beyond current limits. It combines richer family-linked genomic and longitudinal phenotype data, sibling distributions, rare and common variation, functional genomics, and specialised genomic foundation and language models. Current polygenic scores carry real signal. We are building and testing what comes next.
The signal is real. The ceiling is not enough
Polygenic scores for cognitive performance, including IQ-related measures, capture reproducible but still modest signal. Population-level performance can attenuate within families—especially for behavioural and social traits—and shift across ancestry, age, sex and context. Embryo Project treats those limits as measurable design constraints, not a final ceiling.
- 01DataFamily-linked genomic and dense longitudinal phenotype data
- 02ArchitecturePRS ensembles, rare variants, functional biology, and specialised genomic foundation and language models
- 03TestHeld-out families, sibling comparisons, cross-ancestry calibration and prospective follow-up
A model of phenotype beyond a single score
Embryo Project aims to estimate calibrated distributions of possible phenotypes, with uncertainty attached. Every claimed advance must survive held-out family data, sibling comparisons, cross-ancestry calibration and prospective validation.
Current boundary
Educational attainment is not IQ, and IQ is not the mind. Current cognitive scores are reproducible but incomplete; Embryo Project will model general ability, learning, memory, executive function and behavioural traits as distinct outcomes. No genotype-to-adult-phenotype model has yet been prospectively validated for clinical embryo screening, ranking or transfer counselling.
Build the team that can break the ceiling
Statistical genetics, reproductive and developmental biology, longitudinal phenotyping, psychometrics, and genomic foundation and language models. The programme is designed to produce calibrated evidence and make every claimed gain falsifiable.
- 01
Statistical geneticists
- 02
Reproductive and developmental biologists
- 03
Psychometricians and longitudinal phenotype researchers
- 04
Genomic foundation and language-model researchers