How to reproduce a CRISPR-Cas9 knockout from a paper

This guide covers reproducing a published CRISPR-Cas9 knockout, from the guide RNA sequence through delivery to the evidence that the gene is actually gone. Papers often name the target gene and the vector but leave out the protospacer sequence, the exon it cuts, and how clones were confirmed. Those are the details that decide whether your line behaves like theirs.

What methods sections usually leave out

The protospacer sequence and which exon it targets
A gene name does not identify a cut site. Guides in different exons can produce a truncated protein, a hypomorph, or a full loss of function, and the resulting phenotypes differ.
Delivery format and dose
Plasmid, lentivirus, in vitro transcribed RNA, and Cas9 protein with synthetic guide behave differently in editing kinetics, toxicity, and persistence of Cas9 expression.
How the edit was genotyped at the allele level
A mismatch cleavage assay or a gel shift shows that editing happened, but not which indels are present on which alleles. Only sequencing tells you whether every allele carries a frameshift.
Protein-level confirmation
A frameshift can still yield a shortened protein through reinitiation or exon skipping. Without a Western blot or equivalent, a sequenced indel is not proof of protein loss.
Controls for clonal variation
Single-cell cloning itself changes cells. A single knockout clone compared with the parental line conflates editing with clonal drift unless several independent clones or guides are used.
Off-target assessment
Papers frequently state that off-targets were predicted in silico without reporting which sites were checked or sequenced, so you cannot tell whether a phenotype is on-target.

Key parameters and what they change

Parameter	Typical	Changes together with
Guide RNA sequence and position	Typically a 20 nucleotide protospacer next to a 5'-NGG PAM for SpCas9, placed in an early constitutive coding exon	Genotyping primer design, expected amplicon size, and which off-target sites are worth checking.
Cas9 to guide ratio	For ribonucleoprotein delivery, often a modest molar excess of guide over Cas9	Complex volume added per well, editing efficiency, and toxicity.
Cell number per electroporation or transfection	Set by the device or reagent protocol for the plate format in use	Amount of complex or vector, recovery medium volume, and time to the first genotyping.
Selection or sorting window	Antibiotic selection or fluorescence sorting a few days after delivery	When bulk editing is measured and when single-cell cloning can start.
Number of clones screened	Enough to expect several independent biallelic edits, often tens of clones	Number of genotyping reactions, plate layout, and reagent totals for the screen.

Troubleshooting

Symptom	Likely causes	What to check first
Low or no editing in the bulk population	An inefficient guide, poor delivery into the cell type, degraded guide RNA, or a mismatch between the cell line's sequence and the guide.	Confirm delivery with a reporter, sequence the locus in your own cells to rule out polymorphisms at the target site, and test two or three guides in parallel.
Editing in bulk but no biallelic clones	Mostly monoallelic edits, in-frame indels, or loss of the edited cells because the gene is required for growth.	Sequence the amplicon and look at the indel spectrum. If the gene is essential, consider an inducible or conditional approach.
A clone sequences as a frameshift but protein is still detectable	Alternative translation initiation, exon skipping around the indel, or an antibody detecting a related protein.	Blot with antibodies against epitopes both upstream and downstream of the cut, and sequence the transcript to see what is actually produced.
Genotyping gives a messy or mixed sequence trace	The clone is not clonal, or the alleles (often more than two in aneuploid lines) carry different indels.	Deconvolute the trace with an indel analysis tool, or clone the amplicon and sequence individual products; re-single-cell the line if needed.
The knockout phenotype does not match the paper	A different cut site, clonal drift, incomplete knockout, or an off-target effect in one of the lines.	Compare several independent clones from at least two guides, and test rescue by re-expressing the gene.

Worked example: from a methods sentence to executable steps

Knockout cells were generated by transfecting a Cas9 and single guide RNA plasmid targeting the gene of interest, followed by puromycin selection and single-cell cloning. Knockout was confirmed by Sanger sequencing.

Each step is tagged by how clearly the text states it: explicit, partial, inferred, or missing.

1.
Design or obtain the single guide RNA targeting the geneMissing
No protospacer sequence or exon given; check the supplementary tables first.
2.
Transfect the Cas9 and guide plasmidPartial
Vector identity, DNA amount, and transfection reagent are not stated.
3.
Select transfected cells with puromycinPartial
Concentration and duration are not given and are cell line specific.
4.
Expand the population and single-cell cloneExplicit
5.
Amplify the target locus and Sanger sequence clonesExplicit
6.
Number of clones screened and how many were biallelicMissing
Affects whether the reported phenotype came from one clone or several.
7.
Confirm loss of proteinMissing
Sequencing alone does not establish protein loss.
8.
Assess off-target sitesMissing
Not described in the paper; commonly done by predicting and sequencing top sites.

How Vara helps

Paste the methods paragraph and Vara turns it into guide design, delivery, cloning, and validation steps with the parameters it could extract, each tagged by how explicit the source was.
Items the paper never provided, such as the protospacer sequence or the number of clones screened, are listed as reproduction risks instead of being invented.
The screen shows as lanes of sub-experiments with dependencies, so clone expansion and genotyping stay attached to the delivery step they came from.
Reagent amounts scale by formula with cell number and plate format, and each attempt is logged against the protocol version you used.

Try Vara on your own protocol See how importing works

FAQ

How do I reproduce a CRISPR knockout when the paper does not give the guide sequence?+

Check the supplementary tables, the reagent table, and any earlier paper from the same group, since guide sequences are often reported there. If the sequence is still missing, write to the corresponding author. Designing your own guide is reasonable but it is a different cut site, so report it that way.

Why is my editing efficiency low with a guide that worked in the paper?+

Delivery into your cell type is the most common difference. Confirm that the complex or vector actually enters the cells, verify the guide RNA has not degraded, and sequence your cell line at the target site, because a polymorphism in the protospacer or PAM can block cutting.

Is Sanger sequencing enough to confirm a knockout?+

Sequencing shows the indel but not the protein. A frameshift can still produce a shortened product, so confirm loss at the protein level and, where possible, compare several independent clones made with different guides.

How many clones should I screen?+

It depends on editing efficiency in your population. If bulk editing is high, a few dozen clones often yield several biallelic knockouts. Measure bulk editing first, because screening before editing works wastes most of the effort.