Guide RNA selection
Most protospacer sequences used as training data to develop our computational model are derived from human and mouse coding regions, in which the GC content is generally balanced (mainly in the 30% - 70% range) rather than extremely low or high.
Thus, we cannot rule out the possibility that the accuracy our model could be limited in predicting the activity of guide RNAs with extreme GC content such as < 30% or > 70%.
Because guide RNAs with balanced GC content showed the highest activity, we provide, in the web tool, a filter for guide RNA sequences with GC content in the 30 - 70% range as the default setting.
When users turn off this filter, our tool marks sequences with extreme GC content. Guide RNAs with high cindel scores and balanced (30% - 70%) GC content may be chosen as the first options for efficient genome editing. Guide RNAs with high cindel scores and extreme GC content (< 30% or > 70%) or those with low cindel scores and balanced GC content should be regarded as second choices.