Knockout vs Knock-in: The Allele You Want, Not the Enzyme
Gene knockout versus knock-in experimental outcomes are the two genotype results a CRISPR experiment is designed to produce: loss of function at a locus versus a specified insertion or substitution. They are not enzyme names. Cas makes the cut. The allele you want decides whether you usually omit a donor or design one. This page names that intended allele first. The already published NHEJ versus HDR explainer covers the two repair paths after the cut.
Name the Allele, Not the Enzyme
A knockout is an intended inactivation: the locus is meant to stop supplying its normal function. The usual molecular trace is a frameshift or other disrupting indel in a coding exon. The word does not name Cas9, Cas12, or any other nuclease.
A knock-in is an intended sequence change you can write down in advance: a substitution, a tag, a reporter, or another specified insertion. Addgene's CRISPR 101 note on homology-directed repair describes that class of experiment as site-specific knock-in: your Cas enzyme, a guide RNA to the site, and a donor DNA molecule that carries the desired edit. The designed change can still reduce function if that is what the substitution does. The word still names the specified sequence, not the enzyme on the bench.
That is why "we transfected Cas9" is not an experimental description. The same nuclease can serve either goal. Ran and colleagues (2013) built a mammalian Cas9 protocol around editing via NHEJ or via HDR without changing the enzyme. The variable that changes the genotype is the allele you meant — and what you put into the transfection to get it.
Goal, Usual Path, and What You Must Supply
Once the allele is named, the usual mapping onto cellular repair is short. Addgene states that eukaryotic double-strand breaks are repaired primarily by NHEJ or HDR: NHEJ ligates ends and often leaves small insertions or deletions; HDR copies from homologous sequence and can introduce specified changes that exist in the donor. Ran's protocol is the same fork written as a methods paper. Those are the usual paths, not the only published ones. Homology-independent insertion methods exist; they are a different literature and are not the first mapping for this difference query.
| Intended allele | Usual path after the Cas cut | What you must supply |
|---|---|---|
| Loss-of-function knockout | NHEJ, typically leaving inactivating indels at the cut | Cas plus a guide RNA; no donor |
| Specified substitution or insertion (knock-in) | HDR that copies from a homologous donor | Cas, a guide RNA, and a donor designed for that cut |
| Same nuclease, different goal | The path follows the allele, not the brand of Cas | Naming Cas9 does not specify the genotype |
Addgene treats donor class as a function of edit size — short single-stranded oligos for small changes, plasmid or other dsDNA for large cassettes — and warns that Cas will keep cutting if the guide or PAM still matches after a successful edit. Those lengths vary by locus and system. They are constraints, not a recipe, and they are not restated here as universal numbers. Published knock-in frequencies on service pages are likewise locus-specific; they are not planning constants.
Why This Is Not the NHEJ versus HDR Page
The live NHEJ versus HDR page already answers a different question: after Cas cuts both strands, which cellular path is the experiment using, and what cell-cycle and template constraints come with HDR. That page is the repair-path explainer. This page is upstream of it. You name the allele first. You open the path page when you need the fork — template, cell-cycle window, relative speed — not when you still have not said whether you wanted a null or a specified insertion.
Two collisions are worth refusing. First, do not rebuild that path table here. Second, do not treat "knockout versus knock-in" as a synonym for "NHEJ versus HDR." The usual mapping is real, and Addgene and Ran both use it. The names still answer different questions. If a collaborator writes "Cas9 knockout," ask which allele they meant before you ask which path they supplied reagents for.
After the Allele Is Named, Scoring Still Has a Job
Guide-design software answers where this nuclease can cut in a chosen genome and which similar sites look risky. It does not choose the allele. A high on-target score is not evidence that a knock-in will land, and a quiet off-target list is not a knockout setting. After the genotype is named, scoring is still useful: the cut has to exist, and off-target risk is independent of whether you later add a donor.
ZettaCRISPR is one workspace example of that scoring job. The Zettalab product page documents CRISPR guide-RNA design with on- and off-target scoring. That capability nominates guides. It does not choose knockout or knock-in and it does not guarantee an edit. If the remaining decision is which scoring environment the group will sit in, the CRISPOR versus commercial CRISPR software comparison is the next page.
When Loss of Function Is the Complete Goal
A knockout that only needs a disrupted reading frame does not become more rigorous because a donor was added. Donor-free NHEJ is the matching design, not a shortcut. This page stops before homology-independent insertion methods, NHEJ inhibitors, and other HDR-boosting tactics — those papers assume you already knew you wanted a specified allele. It also makes no clinical or therapeutic claim and does not promise that any transfection will yield the intended genotype. Naming the allele is the planning step. Recovering it is still an experiment.
Frequently Asked Questions
Is a knockout just another name for NHEJ?
No. Knockout names the intended loss of function. NHEJ is the usual cellular path that produces the inactivating indels. The path biology lives on the NHEJ versus HDR explainer.
Do I need a donor DNA template for a CRISPR knock-in?
For the usual HDR knock-in, yes. Addgene lists Cas, a guide, and a donor that carries the specified change as the inputs. A typical NHEJ knockout does not use that donor.
Can the same Cas9 enzyme make both a knockout and a knock-in?
Yes. The nuclease makes the break. Ran's 2013 protocol describes editing via NHEJ or HDR with the same Cas9 toolkit. The intended allele — and whether you supply a donor — is what changes the experiment.
Does guide-design software decide knockout versus knock-in?
No. Design tools rank candidate guides for a chosen genome. They do not choose the allele, and a score does not guarantee a knockout or a knock-in. Use software after you know which genotype the experiment is for.
Related Articles