Preventing off target editing by CRISPR-Cas9: machine learning, AI and other techniques of dealing with extreme complexity

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6037314/

https://www.ncbi.nlm.nih.gov/pubmed/30423072

"Last year, CRISPR/Cas9 has been adopted to the gene editing on human embryos (Ma et al., 2017). It raised numerous concerns and public debates. Even worse, the whole-genome sequencing study on a CRISPR/Cas9-edited human and zebrafish cells have revealed an unexpectedly a number of off-target mutations incurred by CRISPR/Cas9 (Fu et al., 2013 and Hruscha et al., 2013). It is extremely necessary for us, human, to manage and control the off-target activities before it becomes too late."

Table 1. Examples of DNA, sgRNA and the coordinates of DNA 
sgRNA                                                      Putative off-target DNA sequences          Chrom  Start          End 
AAATGAGAAGAAGAGGCACAGGG AAATGAGAAGAAGAGGCACAGGG Chr3    46399717  46399739 
AACACCAGTGAGTAGAGCGGAGG  TACTCCAGTGAGTAGAGAGGCGG   Chr16  87942783  87942805 
CTTGCCCCACAGGGCAGTAACGG   TCAGCCCCACAGGGCAGTAAGGG   Chr9   104595866 104595888 
GAACACAAAGCATAGACTGCGGG   GAACACAATGCATAGATTGCCGG    Chr4   90522167    90522189 
GACACCGAAGCAGAGTTTTTAGG   GACAACAAAGTAGAGTTTTTAGG   Chr7    100221100  100221122 

In each DNA sequence, the last three nucleotides are PAM highlighted in bold.