TABLE 1.

In silico and experimental methods for genome-wide off-target prediction.

		Methods	Characteristics	Advantages	Disadvantages
In silico prediction	Alignment based models	CasOT [21]	Adjustable in PAM sequence and the mismatch number (at most 6)	Conveniently accessable via internet	Biased toward sgRNA-dependent off-target effects; results need experimental validation
		Cas-OFFinder [22]	Adjustable in sgRNA length, PAM type, and number of mismatches or bulges
		FlashFry [23]	Provides information about GC contents
		Crisflash [24]	High in speed
	Scoring based models	MIT [15, 25]	Based on the position of the mismatches to the gRNA
		CCTop [26]	Based on the distances of the mismatches to the PAM
		CROP-IT [27]
		CFD [28]	Based on a experimentally validated dataset
		DeepCRISPR [29]	Considers both sequence and epigenetic feature
		Elevation [30]
Experimental detection	Cell-free methods	Digenome-seq [31–33]	Digests purified DNA with Cas9/gRNA RNP → WGS	Highly sensitive	Expensive; requires high sequencing coverage; requires a reference genome
		DIG-seq [34]	Uses cell-free chromatin with Digenome-seq pipeline	Concerning chromatin accessibility; higher validation rate than Digenome-seq
		Extru-seq [35]	Pre-incubates live cells with Cas9/sgRNA RNP complex→rapidly kill cells by extruder→WGS	Low miss rate; low false positive rate	Expensive; difficult to detect Cas9-mediated large deletions, chromosomal depletions, and translocations
		SITE-seq [37]	A biochemical method with selective biotinylation and enrichment of fragments after Cas9/gRNA digestion	Minimal read depth; eliminated background; does not require a reference genome	Low sensitivity; low validation rate
		CIRCLE-seq [38–40]	Circularizes sheared genomic DNA→incubate with Cas9/gRNA RNP→linearized DNA for NGS
	Cell culture-based methods	WGS [41–43]	Sequences the whole genome before and after gene editing	Comprehensive analysis of the whole genome	Expensive; limited number of clones can be analyzed
		ChIP-seq [44–47]	Analyzes binding sites of catalytically inactive dCas9	Detection of Cas9 binding sites genome-wide	Low validation rate; affected by antibody specificity and chromatin accessibility
		IDLV [48–52]	Integrates IDLV into DSBs	Detects off-targets in cells that are difficult to transfect	Low sensitivity; high false positive rate
		GUIDE-seq [36, 53–55]	Integrates dsODNs into DSBs	Highly sensitive, low in cost, low false positive rate	Limited by transfection efficiency
		LAM–HTGTS [57–59]	Detects DSB-caused chromosomal translocations by sequencing bait-prey DSB junctions	Accurately detects chromosomal translocations induced by DSBs	Only detects DSBs with translocation; efficiency limited by chromatin accessibility
		BLESS [60, 61]	Captures DSBs in situ by biotinylated adaptors	Directly capture DSBs in situ	Only identifies off-target sites at the time of detection
		BLISS [61, 62]	Captures DSBs in situ by dsODNs with T7 promoter sequence	Directly capture DSBs in situ; low-input needed
	In vivo detection	Discover-seq [63]	Utilizes DNA repair protein MRE11 as bait to perform ChIP-seq	Highly sensitive; high precision in cells	Has false positives
		GUIDE-tag [64]	Uses biotin-dsDNA to mark DSBs	Highly sensitive; detects off target sites in vivo	The incorporation rate of biotin-dsDNA is relatively low (∼6%)