Design of guide RNAs for genome editing of Yarrowia lipolytica using deep learning

The yeast Yarrowia lipolytica is capable of producing important natural and recombinantproducts with industrial and therapeutic commercial value. The CRISPR/Cas system is used for genomeediting of this yeast to achieve effective metabolic engineering and increase the production efficiencyof valuable products. In this system, a protein called Cas combines with a short RNA called guide RNA(sgRNA) and makes a double-strand break precisely at the desired location in the genome. If the sgRNAis not designed with high precision and without off-target, unexpected off-target edits can occur andpotentially cause harmful effects. Hence, the aim of the current study is to design precise and less offtargetsgRNA by computational and deep learning [۱]. In the new computational model, with the helpof new functions and articles, chemical and physical features were extracted for each sequence.Influential features were identified using an encoder and decoder, and a model was obtained using CNN.With the help of cutting score (CS) that Baisya et al. [۲] obtained in the laboratory for each sgRNA andusing computational learning models. They obtained Spearman value of ۰.۳۷% and Pearson value of۰.۴۳%. In the current study, Spearman value and Pearson value reached to ۰.۴۴% and ۰.۵۲%,respectively. Furthermore, a library was designed to produce sgRNA sequences for the Cas۹ proteinusing the alignment method for the genome of each organism so that the effectiveness of the sequencescould be evaluated with the sequence obtained in the laboratory. A computational model was alsopresented to obtain the epigenetic value for each sgRNA with the least off-target. The findings can leadto a better and more accurate design of sgRNA with the least off-target which will greatly help thegenome engineering of yeasts and other organisms.

کلیدواژه ها:

Guide RNA ، Machine learning ، Deep learning ، Yeast ، Genome editing