With the first sequence of a human genome nearing completion at the turn of the 21st century, many scientists predicted they would be able to pinpoint individual genes responsible for diseases within a matter of years.

At a 2000 press conference announcing the first draft of the Human Genome Project, President Bill Clinton remarked, “In coming years, doctors increasingly will be able to cure diseases like Alzheimer’s, Parkinson’s, diabetes, and cancer by attacking their genetic roots.”

More than 25 years later, that sentiment strikes many scientists as a bit naïve. It turns out that our 20,000 (give or take) genes rarely work alone. The more scientists plumb the human genome to find mutations linked to a given disease, the more they find. Dozens of genes have since been linked to Alzheimer’s and Parkinson’s diseases, and hundreds, if not thousands, to Type 2 diabetes and cancer.

Scientists have begun to realize that the genetic roots of disease – or of nearly any other bodily trait – lie not in individual genes, but in what are called gene regulatory networks, large intermeshed systems of genetic interactions that result in a given output like a disease. That makes developing new therapies a challenge because researchers often don’t know which of the many parts of the network to target.

Recently, a team of researchers from Stanford Medicine and the Gladstone Institutes debuted a massive dataset that could help make sense of those large, tangled gene networks. The dataset, described in an article published in Cell, captures what happens to the activity of all the genes in human immune cells, known as T cells, when an individual gene is dialed down in those cells.

A lookup table for disease

The research team, led by Gladstone’s Alex Marson, MD, PhD, and Stanford Medicine’s Jonathan Pritchard, PhD, the Bing Professor in Population Studies and a professor of genetics and of biology, used a technique known as perturb-seq (so-called because the method involves perturbing one aspect of the genome and then sequencing large numbers of other genetic material in the cell) to methodically dial down the activity of 12,748 genes in 22 million human T cells from four different donors. They used the gene-editing tool CRISPR to dial them down one at a time and capture the resulting changes in the activity of all the cells’ other genes. The publicly available dataset sheds new light on how gene regulatory networks operate in healthy human T cells, but what these gene activity patterns tell us about disease will have immense value to medical research, Pritchard said.

Other research groups have made and published many different datasets profiling how gene activity changes across the human genome in diseased cells. Much like scientists have found dozens to thousands of genetic mutations linked to human disease, when they look at gene activity in diseased versus healthy cells, they see changes in hundreds to thousands of different genes. It may be impossible to know which are promising targets for new therapies.

But now, scientists can use the perturb-seq dataset as a type of lookup table, matching the gene activity pattern seen in a given disease with one that results from dialing down the activity of a single gene in healthy cells. If they find a pattern match, that gives the scientists a clue that the dialed-down gene from the perturb-seq dataset might be especially important in that disease — and could be a target for new therapies.

“Unless you know the network, none of this makes sense,” Pritchard said. “The genetic data is really confusing. With the perturb-seq data, you’re measuring something different, but it gives you this tool for interpreting that confusing data.”

To prove the translating power of their dataset, the scientists used it to make sense of two existing genetic datasets: one that catalogs genes connected to a variety of autoimmune diseases, and one that profiles rare mutations associated with having too many or too few immune cells. In both cases, they found matching patterns in their perturb-seq data. Their dataset is freely available to anyone else who wants to use it; Pritchard said he’s already heard of several groups who are using it in their research.

The first dataset of its kind

The scientists chose T cells because they are pivotal for pretty much every function of the immune system, including fighting off infections, recognizing and eliminating pre-cancerous cells, and wound healing. They also play a role in most diseases, including cancer and autoimmune disease. And they form the basis for an important form of immunotherapy known as CAR-T cell therapy, in which a patient’s own T cells are engineered in the lab to recognize and eliminate cancerous cells or autoimmunity-triggering cells.

Although other research groups have released similar perturb-seq datasets with other types of cells, this is the first time the technique has been used at such a large scale on what are known as primary cells, cells taken directly from human donors. Other large perturb-seq datasets used human cell lines, typically cancer cells that have been cultured in the lab for many generations and may not accurately represent true human biology in all cases. That required new methods, developed in the Marson lab, to get the CRISPR-based technique working in human T cells.

“It’s the first time anybody’s done this type of experiment at genome scale and in a primary cell type,” Pritchard said. “It’s a milestone in terms of the technology development.”

The researchers are working on a similar perturb-seq dataset in which they increase the activity of those same genes, rather than dialing them down. They’re building another dataset with a different type of immune cell called regulatory T cells, and Pritchard’s group is developing new statistical methods to help scientists compare different large datasets, to better mine the perturb-seq data for new insights about health and disease.

“This is really a tool that you can point at a lot of different problems,” Pritchard said.