The ICE algorithm was developed internally by Synthego to analyze editing results using standard Sanger sequence reads. It is different from, but yields outputs that are similar to, the Tracking of Indels by Decomposition (TIDE) analysis, but with important extra features.
The ICE modeling process reports an ICE score and a Pearson regression coefficient (r2). If the ICE r2 is high, the observed data is a very good fit with a combination of proposed edited sequences and you can be confident in the reported editing efficiencies. An ICE score of 100 with a high r2 value, for example, indicates that there is no wild-type sequence in the cell population being interrogated. If the r2 of an ICE score is low, the observed data fits poorly to any combination of proposed edited sequences. In that case, we recommend looking at the ICE-D score.
ICE-D can capture larger indels, such as multiplex editing events that utilize several sgRNAs to target and cut a specific locus, larger insertions and deletions. ICE-D will not specifically detail these complex sequence patterns, but will determine the quantity of wild-type sequence that is present in a cell population. Given ICE-D’s ability to identify a larger range of editing outcomes, it is possible that ICE and ICE-D values will not be equal.