Skip to main content
Premium Trial:

Request an Annual Quote

Combined Genomic Analyses Begin to Shed Light on Disease Causality

COLD SPRING HARBOR, NY (GenomeWeb) – New genomic tools can help researchers begin to tease out disease causality, CeMM Research Center for Molecular Medicine's Christoph Bock said during a talk at the Biology of Genomes here.

Approaches like single-cell RNA-sequencing and ATAC-seq (assay for transposase-accessible chromatin using sequencing) give researchers insight into what occurs during disease development, he added, but they cannot establish causality. Instead, Bock noted that there are four general approaches to establish causality.

The first way is a biochemical approach — which Bock dubbed "one gene, one postdoc" — though he noted that it is also a conservative method. Another is to rely on Mendelian randomization. But this approach requires a lot of data, which Bock said might not yet be available.

Instead, Bock said he has focused on a third and a fourth approach, using time series and perturbation approaches to examine causality. A time series relies on Granger causality, which is useful but limited in its formal proof, he cautioned. CRISPR-based perturbation and single-cell sequencing, he added, could uncover functional evidence at scale.

For a time series analysis, Bock turned to chronic lymphocytic leukemia as a model disease. People have a 1 in 200 lifetime risk of developing CLL, which is typically diagnosed later in life, at an average of about 70 years.

As he and his colleagues previously reported in Nature Communications, they conducted an ATAC-seq analysis of 88 CLL samples from 55 patients. Their resulting genome-wide chromatin accessibility maps showed a strong separation between two disease subtypes, IGHV-unmutated uCLL and IGHV-mutated mCLL.

Following on that, they have since used a combination of ATAC-seq, single-cell RNA-seq, and cellular phenotyping to gauge CLL response to ibrutinib, a Bruton tyrosine kinase inhibitor, over eight different time points in seven patients. From this, Bock said they were able to reconstruct the order of events that occurred in patients in response to treatment.

In particular, they found that the same regulatory program was consistently shared across patients. In it, there was a decrease in NF-κB binding followed by reduced regulatory activity of lineage-linked factors like PAX5 and IRF4, and then the loss of CLL cell identity. But they also noted a difference between patients in how quickly this occurred. A preprint describing this work was posted to BioRxiv last month.

Ultimately, Bock said they want to be able to do patient-specific route planning — to know where the disease is heading before it gets there with the goal of delaying disease.

Meanwhile, on the perturbation side, Bock noted CRISPR could be repurposed for high-throughput editing to help tease out causality. He and his colleagues developed an approach they dubbed CRISPR droplet sequencing, or CROP-seq, that combines CRISPR-Cas9 screening and single-cell RNA-seq.

In particular, this approach marries the benefits of pooled screens with those of arrayed screens. Pooled CRISPR screens, he noted, work well for strong phenotypes. But they don't support complex molecular readouts. Arrayed CRISPR screens do allow for such readouts, but have a limited throughput, he noted.

"We need an assay to read effects out in high throughput," he said.

To do so, CROP-seq uses a CRISPR guide RNA as a barcode that is later expressed and detected through transcriptome sequencing, as well as a high-throughput assay for single-cell RNA-seq, a computational approach to assign those single-cell transcriptomes to gRNAs, and a means of analyzing and interpreting gRNA-induced transcriptional profiles. They then validated this approach on T cell receptor induction.

Since developing the tool, Bock said that they've been working to improve it and scale it up. He added that CROP-seq could be combined with any single-cell RNA-seq approach or multi-omic assays, and could be scaled to genome-wide assays and applied to ex vivo and in vivo screens.

More than 200 other labs have tried this approach, he said, noting that it is an open-source protocol.