Deep Learning Model Accurately Predicts RNA-Targeting CRISPR Activity

According to new research published in Nature Biotechnology, artificial intelligence can predict on- and off-target behavior of CRISPR tools that target RNA rather than DNA.

Deep Learning Model Accurately Predicts RNA-Targeting CRISPR Activity

Image Credit: New York University

The study by researchers at the New York Genome Center, Columbia University, and New York University uses a deep learning model and CRISPR screens to control the expression of human genes in various ways, similar to turning a light switch to completely turn them off or turning a dimmer knob to partially reduce their activity.

New CRISPR-based treatments could be created using these precise gene controls.

CRISPR is a gene-editing technology with numerous applications in biomedicine and elsewhere, ranging from treating sickle cell anemia to creating mustard greens with improved flavor. It frequently functions by utilizing the Cas9 enzyme to target DNA. Recent years have seen the discovery of a different kind of CRISPR that uses the Cas13 enzyme to target RNA.

RNA editing, knocking down RNA to stop the expression of a specific gene, and high-throughput screening to find promising drug candidates are just a few uses for RNA-targeting CRISPRs.

To better understand RNA regulation and determine the role of non-coding RNAs, researchers at NYU and the New York Genome Center developed a platform for RNA-targeting CRISPR screens utilizing Cas13.

Since RNA serves as the primary genetic component of viruses like SARS-CoV-2 and the flu, RNA-targeting CRISPRs have the potential to contribute to the development of fresh approaches for the prevention or treatment of viral diseases.

Additionally, the production of RNA from the genome’s DNA occurs as one of the first stages in human cells when a gene is expressed.

The study’s main objective is to increase RNA-targeting CRISPR activity on the chosen target RNA and decrease activity on other RNAs that could have negative side effects for the cell. Mismatches between the target RNA and the guide RNA, as well as insertional and deletion mutations, are examples of off-target activity.

Predicting off-target activity, particularly insertion and deletion mutations, has not been thoroughly examined; earlier studies of RNA-targeting CRISPRs primarily focused on on-target activity and mismatches.

Insertions or deletions account for around one in five mutations in human populations, making them significant categories of possible off-targets to take into account while developing CRISPR strategies.

Similar to DNA-targeting CRISPRs such as Cas9, we anticipate that RNA-targeting CRISPRs such as Cas13 will have an outsized impact in molecular biology and biomedical applications in the coming years. Accurate guide prediction and off-target identification will be of immense value for this newly developing field and therapeutics.

Neville Sanjana, Study Co-Senior Author and Assistant Professor, Biology, New York University

Sanjana and his colleagues ran a number of pooled RNA-targeting CRISPR screens in human cells for their findings published in Nature Biotechnology. A total of 200,000 guide RNAs, including “perfect match” guide RNAs as well as guide RNAs with off-target mismatches, insertions, and deletions, were evaluated for activity in human cells.

Sanjana’s lab collaborated with David Knowles’ lab to develop TIGER (Targeted Inhibition of Gene Expression by guide RNA design), a deep learning model that was trained using the results of the CRISPR screens.

TIGER outperformed earlier models created for Cas13 on-target guide design and provided the first tool for predicting off-target activity of RNA-targeting CRISPRs when comparing the predictions produced by the deep learning model and laboratory tests in human cells.

Machine learning and deep learning are showing their strength in genomics because they can take advantage of the huge datasets that can now be generated by modern high-throughput experiments. Importantly, we were also able to use “interpretable machine learning” to understand why the model predicts that a specific guide will work well.

David Knowles, Study Co-Senior Author and Assistant Professor, Computer Science and Systems Biology, School of Engineering and Applied Science, Columbia University

Hans-Hermann (Harm) Wessels, the study’s co-first author and a senior scientist at the New York Genome Center, who was previously a postdoctoral fellow in Sanjana’s laboratory, added, “Our earlier research demonstrated how to design Cas13 guides that can knock down a particular RNA. With TIGER, we can now design Cas13 guides that strike a balance between on-target knockdown and avoiding off-target activity.

By permitting partial suppression of gene expression in cells with mismatch guides, the researchers also showed that TIGER’s off-target predictions could be used to accurately regulate gene dosage—the quantity of a particular gene that is expressed.

This may be helpful for conditions where there are too many copies of a gene, such as Down syndrome, specific types of schizophrenia, Charcot-Marie-Tooth disease (a hereditary nerve disorder), or for diseases like cancer in which aberrant gene expression can cause unchecked tumor growth.

Our deep learning model can tell us not only how to design a guide RNA that knocks down a transcript completely, but can also ‘tune’ it—for instance, having it produce only 70% of the transcript of a specific gene.

Andrew Stirn, Study Co-First Author and PhD Student, Columbia Engineering and New York Genome Center

The researchers hope that by combining artificial intelligence with an RNA-targeting CRISPR screen, TIGER’s predictions would help prevent undesirable off-target CRISPR activity and accelerate the creation of a new generation of RNA-targeting therapies.

Sanjana stated, “As we collect larger datasets from CRISPR screens, the opportunities to apply sophisticated machine learning models are growingly rapid. We are lucky to have David’s lab next door to ours to facilitate this wonderful, cross-disciplinary collaboration. And, with TIGER, we can predict off-targets and precisely modulate gene dosage which enables many exciting new applications for RNA-targeting CRISPRs for biomedicine.

Building on the NYU team’s earlier work to create guide RNA design rules, target RNAs in a variety of organisms, including viruses like SARS-CoV-2, engineer protein and RNA therapeutics, and use single-cell biology to uncover synergistic drug combinations for leukemia, this most recent study expands the broad applicability of RNA-targeting CRISPRs for human genetics and drug discovery.

Journal reference:

Wessels, H.-H., et al. (2023). Prediction of on-target and off-target activity of CRISPR–Cas13d guide RNAs using deep learning. Nature Biotechnology.


The opinions expressed here are the views of the writer and do not necessarily reflect the views and opinions of AZoLifeSciences.
Post a new comment

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.

You might also like...
Unleashing the Potential of CRISPR-COPIES