Scientists Unite Quantum Biology & AI for Gene Editing Accuracy

ORNL scientists used quantum biology, AI, and bioengineering to improve CRISPR Cas9 genome editing for microbes, boosting renewable fuel and chemical production.

CRISPR is a powerful bioengineering tool, that alters genetic code for improved organism performance or mutation correction. CRISPR Cas9 targets and cleaves specific genome sites using a single guide RNA. Previous predictive models for effective guide RNAs lacked accuracy in microbes due to limited data from specific species.

Mammalian cells

Carrie Eckert from ORNL’s Synthetic Biology group noted that CRISPR tools, primarily designed for model species, show gaps in adapting to microbes with distinct chromosomal structures and sizes, a finding confirmed by this study.

Therefore, ORNL researchers improved guide RNA design by investigating the processes within cell nuclei where genetic material resides. They used quantum biology, merging molecular biology and quantum chemistry, to understand how electronic structure influences the properties of nucleotides, the building blocks of DNA and RNA.

“The distribution of electrons within the molecule impacts reactivity and structural stability, affecting the ability of the Cas9 enzyme-guide RNA complex to bind effectively with the microbe’s DNA,” noted Erica Prates, computational systems biologist at ORNL.

Scientists unite quantum biology & AI for gene editing accuracy: The best guide through a forest of decisions

The scientists devised an interpretable artificial intelligence model called iterative random forest, described in the journal Nucleic Acids Research, employing around 50,000 guide RNAs designed for E. coli bacteria’s genome. This model integrated quantum chemical properties, aiming to enhance guide RNA selection.

Therefore, this approach unveiled crucial nucleotide features that contribute to better guiding RNA selection. “Through this model, we gleaned insights into the molecular mechanisms governing guide RNA efficiency,” highlighted Prates. “It’s essentially a rich molecular database aiding in advancing CRISPR technology.”

The ORNL team verified the explainable AI model’s accuracy by performing CRISPR Cas9 cutting experiments on E. coli, employing a large set of guides chosen by the model.

The utilization of explainable AI provided a deeper understanding of the biological processes behind the outcomes. So, Jaclyn Noshay, the paper’s first author and a former ORNL computational systems biologist, emphasized the contrast with deep learning models, which often operate as “black box” algorithms, lacking interpretability.

“We aimed to refine our grasp of guide design principles to enhance cutting efficiency, particularly in microbial species, recognizing the limitations of models trained across different biological kingdoms,” Noshay emphasized.

With thousands of features, the AI model was trained using ORNL’s Summit supercomputer at the Oak Ridge Leadership Computer Facility (OLCF), a DOE Office of Science user facility.

Scientists unite quantum biology & AI for gene editing accuracy: quantum properties

Moreover, Eckert expressed intentions to collaborate with her synthetic biology team and computational science counterparts at ORNL, aiming to further enhance the newly developed microbial CRISPR Cas9 model by leveraging experimental data or exploring various microbial species.

Therefore, considering quantum properties offers the potential for improving Cas9 guides applicable to various species. Therefore,” this study extends its implications even to human-scale applications,” Eckert noted. “Accurate guide predictions are crucial, whether it’s for drug development targeting specific genome regions.”

Enhancing CRISPR Cas9 models streamlines the process of linking genotype to phenotype, advancing functional genomics. This research is significant for projects like the ORNL-led Center for Bioenergy Innovation (CBI), which focuses on enhancing bioenergy feedstock plants and bacterial biomass fermentation.

“We aim to significantly enhance the predictive capacity of guide RNA with this research,” Eckert highlighted. “A deeper understanding of biological mechanisms and more extensive data inputs will refine our targeting, amplifying precision and research velocity.”

So, ORNL’s Paul Abraham, overseeing the DOE Genomic Science Program’s SEED SFA backing the CRISPR research, hailed the study’s progress: “This marks a significant stride in understanding how to avoid errors in an organism’s genetic code. I look forward to refining these predictions with more data and ongoing use of explainable AI modeling.”

Read the original article on ScienceDaily.

Read more: Bioactive Ingredients a Quite New Concept.