Our science

The question

Mutations in the epigenetic machinery turn up again and again in neurodevelopmental and psychiatric disorders. They rarely break a single pathway. They shift the probability that a cell takes one developmental route rather than another.

The approach

We take patient-derived and CRISPR-engineered lines through differentiation and profile them at single-cell resolution, so we can look for the point at which trajectories diverge rather than only the difference at the end.

A practical question, not only a biological one

Much of the field assumes that engineering a mutation into a healthy reference line is the cleaner experiment, because everything else is held constant. That assumption is worth testing directly.

What we compare

Patient-derived lines carrying a variant, isogenic engineered lines carrying the same variant, and wild-type controls, differentiated side by side across several genetic backgrounds.

The cohort

Molecular data from stem cell disease models across 210 individuals with rare genetic diseases, building on the Human Induced Pluripotent Stem Cell Initiative.

The question

Before any differentiation is attempted, is a disease genotype already visible in the transcriptome of an undirected cell? The answer determines how much of disease modelling has to happen downstream.

From population to cell

Association studies tell us where in the genome a signal sits. They do not tell us which cell type it acts in, or when. This topic exists to carry those signals into the cellular models the rest of the group builds.

Same cell, two measurements

Morphology and expression are usually measured in different experiments and compared as averages. Measuring both in the same cells turns a correlation between populations into a relationship between observations.

A single gene as a way in

Following one gene from a loss-of-function variant through to a disrupted regulatory programme gives a complete chain of reasoning that broader studies can then be checked against.