AI modifies microbial genes to produce renewable fuels

For over half a decade, the Centre for Bioenergy Innovation (CBI) has been tinkering with plant and microbial genes to address the ultimate challenge of a sustainable, bio-based economy.

The CBI uses supercomputers, deep learning algorithms and gene editing tools such as CRISPR Cas9 to link genotype to phenotype. However, their research has been plagued by the incompatibility of existing gene-splicing models with microbial genomes.

To improve the design of CRISPR Cas9 machinery, scientists at Oak Ridge National Library (ORNL) built an explainable AI model and trained it to identify the molecular cues that enable targeting of specific regions of a microbial genome. The new model, called an iterative random forest, has already been shown to optimize the cleavage and binding of DNA into E. Coli bacteria.