Traditional CRISPR-Cas9 uses a guide RNA to find the target sequence, then Cas9's two nuclease domains cut both DNA strands at once. The cell must then stitch the break back together itself — usually via Non-Homologous End Joining, which is fast but sloppy: it reconnects the ends correctly only some of the time, and otherwise leaves behind small random insertions/deletions, or occasionally a larger structural glitch near the cut site.
Base editing fuses a catalytically disabled Cas9 to a deaminase enzyme. It still uses a guide RNA to find the target, but it never cuts the backbone at all — it reaches into the open strand and directly converts one base's chemical structure into another (e.g. cytosine → thymine) at a precise position in the editing window. Because there is no break to repair, the outcome is far more consistent, though a nearby "bystander" base in the window can occasionally be converted too.
Traditional: bind → cut both strands → NHEJ repair → variable outcome
Base editing: bind → direct chemical conversion → no break → consistent outcome
- Attempt edit — runs one full animated cycle of the currently selected method on the target base pair.
- Run 20× — instantly resolves 20 more attempts to build up the outcome statistics faster.
- Auto-run — repeats attempts automatically so you can watch the distribution settle.
- Switch methods any time — both methods keep independent running totals so you can compare them directly in the chart above.
This precision/predictability gap is why base editors are being explored for point mutations where an unpredictable indel would be worse than no edit at all, while traditional double-strand-cut CRISPR remains well suited to disrupting or inserting larger sequences.