In silico performance analysis of web tools for CRISPRa sgRNA design in human genes

Abstract

Angiogenic gene overexpression has been the main strategy in numerous vascular regenerative gene therapy projects. However, most have failed in clinical trials. CRISPRa technology enhances gene overexpression levels based on the identification of sgRNAs with maximum efficiency and safety. CRISPick and CHOP CHOP are the most widely used web tools for the prediction of sgRNAs. The objective of our study was to analyze the performance of both platforms for the sgRNA design to angiogenic genes (VEGFA, KDR, EPO, HIF-1A, HGF, FGF, PGF, FGF1) involving different human reference genomes (GRCH 37 and GRCH 38). The top 20 ranked sgRNAs proposed by the two tools were analyzed in different aspects. No significant differences were found on the DNA curvature associated with the sgRNA binding sites but the sgRNA predicted on-target efficiency was significantly greater when CRISPick was used. Moreover, the mean ranking variation was greater for the same platform in EPO, EGF, HIF-1A, PGF and HGF, whereas it did not reach statistical significance in KDR, FGF-1 and VEGFA. The rearrangement analysis of the ranking positions was also different between platforms. CRISPick proved to be more accurate in establishing the best sgRNAs in relation to a more complete genome, whereas CHOP CHOP showed a narrower classification reordering.