Our science

Reading the letters, words, and sentences that build the heart.

Almost every heart disease traces back to our DNA. But the genome is a book of three billion letters, and we are only beginning to read the parts that decide when each gene turns on. This is the story of how we map that hidden language — and use it to find cures.

Scroll the story

01The genome

Three billion letters. Barely any of them are genes.

Your genome is a book of 3 billion letters. Only about 2% spell out genes — the proteins that do the work of the cell. The other 98% was long dismissed as junk. It is not junk. It holds the instructions for when and where each gene turns on.

02Enhancers

Hidden in between: two million switches.

Scattered through the noncoding genome are roughly two million enhancers — short stretches of DNA that switch nearby genes on. Each cell type reads a different set of these switches to become what it is. They are the words we are only beginning to understand.

03Variant to function

This is where disease hides.

More than 100,000 disease-linked genetic variants fall inside these switches — including hundreds tied to coronary artery disease, the leading cause of heart attack, and thousands linked to congenital heart disease, which affects about 1% of newborns. The hard question: which gene does each variant actually break?

04CRISPR + Perturb-seq

So we break them — 2,000 at a time.

It has been hard to study one switch at a time, let alone thousands. Helen Kang and Gavin Schnitzler in the lab built a way to break ~2,000 candidate switches at once with CRISPR, then read the consequences in millions of single cells. Understanding heart disease, 2,000 words at a time.

05The wiring

We map each switch to the gene it controls.

Using massively parallel CRISPR screens and our Activity-by-Contact model, we trace each enhancer to its target gene — in each cell type. The wiring rewires from one heart cell type to the next. Switch cell types below to watch it move.

06The payoff

~100 switches that tune your risk of a heart attack.

Reading out endothelial cells — the lining of your blood vessels — the screen surfaced about 100 switches that control heart-attack risk. Each is a possible handle for a new therapy. People carrying changes in these words are more likely to have a heart attack, and, encouragingly, more likely to respond well to statins.

Our mission

Three moonshots for the heart.

The lab and the BASE Initiative at Stanford are organized around three long-horizon goals — from understanding disease, to correcting it, to rebuilding the heart itself.

01
Understand why

Read the letters, words, and sentences that build the heart.

Defects in the heart come from defects in our DNA — but the genome is a big place, and we do not yet know which letters are responsible. We are building a genetic map of the heart to find the sequences that construct it, so we can find where they go wrong. One day, we want to tell every family exactly why their child has a heart defect.
02
Gene-editing therapies

Correct the DNA defect at its source with CRISPR.

A lasting cure has to fix not just the heart structure, but the underlying DNA. Once we know where the defect is, we have the tool to correct it. The first CRISPR therapy was approved just last decade, for sickle cell disease. We are working to design the first CRISPR therapies for children with heart disease.
03
Build a new heart

Reprogram the genome and bioengineer new tissue.

Some defects form so early in development that CRISPR alone cannot fix them. Here we work with surgeons and bioengineers to reprogram the genome — coaxing cells to form all of the dozens of cell types in the heart — and to 3D-print new blood vessels, valves, and heart tissue.

How we work

Four ways of seeing the same problem.

CRISPR genomics

Massively parallel CRISPR screens — CRISPRi, base editing, and Perturb-seq — that break thousands of regulatory switches at once and read out the consequences in single cells.

AI & computation

Deep-learning models that learn the DNA code cell type by cell type, plus the Activity-by-Contact model that predicts which enhancer controls which gene.

Human genetics

More than 100,000 disease-linked variants, connected to the genes, cell types, and pathways they act through — turning statistical associations into biological mechanism.

Team science

Genetics, computer science, bioengineering, and biology under one roof — and a consortium spanning eight institutions applying these maps to human disease.

The science, in depth

What we are actually building.

01 ABC model

Mapping the regulatory wiring of the genome

Enhancers often regulate multiple genes across long distances, one gene can have many enhancers, and these connections rewire across cell types. Our Activity-by-Contact model explains much of this complexity — a strategy to build comprehensive maps of enhancer–gene connections across hundreds of cell types.

How do enhancer–gene connections rewire across cell types, states, and trajectories?
Can we find new mechanisms by studying the elements that break the ABC rule?

02 CRISPR technology

New tools to perturb the noncoding genome

The last decade of genomics was driven by our ability to observe genomic processes — RNA, chromatin state, 3D architecture. The next decade is about perturbing them. We pair CRISPR tools with novel single-cell readouts to run screens across the noncoding genome.

Can we break thousands of enhancers to learn how they find their target genes?
Can we edit thousands of variants to learn the sequence rules of the regulatory code?

03 Variant to function

Connecting variants to functions to understand disease

Genetic studies have linked >100,000 variants to human traits — including >300 for coronary artery disease and thousands of candidates for congenital heart disease. Each could point to a new mechanism, if only we could connect it to a gene, cell type, and pathway. That is the V2F Initiative.

Which genes and pathways in vascular cells control risk for heart disease?
How does the noncoding genome contribute to congenital heart disease?

Explore our data

The maps are open. Use them.

Our maps are catalytic for researchers worldwide. We lead the NIH-funded IGVF consortium — a team nucleated across eight institutions using these maps to understand the genetics of disease.

E2G — Enhancer-to-Gene browser Explore predicted enhancer–gene connections across cell types. IGVF Data Portal Open data from the Impact of Genomic Variation on Function consortium. EngreitzLab on GitHub The Activity-by-Contact model and analysis pipelines, open source.

Read the publications Support the work