Genome editing, targeting and unintended effects

Targeted change is an aim rather than a complete result

Genome editing aims to change genetic information at a selected location. Depending on the approach, the intended change can concern a base, an insertion, a deletion or another sequence relationship. The target identifies where a change is intended; it does not establish what actually resulted or which other properties changed. A resulting genotype and its biological consequences remain different levels of the evidence argument.

The familiar scissors analogy describes approaches involving DNA cleavage and subsequent cellular repair. It is not a universal definition of editing. Base-editing research established approaches that change a target base without requiring a double-strand DNA break. Different mechanisms have different scopes, so neither “all editing cuts both strands” nor “editing always produces the exact intended result” is a sound generalisation.

Unintended outcomes can occur at the selected site as well as elsewhere. An intended location does not ensure a particular repaired sequence, and an otherwise desired target result does not rule out another change. The relevant evidence therefore includes actual outcomes and what was examined, rather than relying on the name of a tool or the specificity of the original intention.

Count desired and elsewhere outcomes on separate axes

The original complete matrix assigns two Boolean attributes to each of twenty records: desired target outcome and the specified elsewhere change. Twelve have the desired outcome only, two have both attributes, four have neither and two have the elsewhere change only. These are exhaustive categories for this constructed account, not a claim that all real biological effects have been examined.

Desired outcomes therefore total fourteen, while elsewhere changes total four. The two counts overlap in two records. Adding fourteen and four would double-count those records unless the overlap were subtracted. Describing desired outcomes as 14/20 and desired-without-the-assigned-elsewhere-change as 12/20 also answers different questions. One success label cannot serve both meanings.

The attributes are separately specified, not stipulated statistically independent. Their joint occurrence is explicitly supplied. Nor does a negative elsewhere attribute establish the absence of every possible unwanted effect: the matrix includes only its named attribute. Biological severity, function, persistence and wider consequences are not encoded by the Boolean counts.

What was untested remains unknown

The separate coverage example knows every desired-target result but examines the elsewhere attribute in only eight identities. One tested result is positive and seven are negative; the remaining twelve identities have no elsewhere result. This is a different evidence situation from the complete matrix. The missing entries cannot be filled using that other model’s counts or by assuming that no record means no change.

The tested fraction 1/8 belongs to the supplied tested group. Selection of those identities was not justified as random or representative, so it does not establish a whole-case frequency. With twelve unknown entries, the total elsewhere-positive count can still range from one to thirteen. This is a range of compatible inventories, not a probability distribution or an estimate of biological hazard.

Claims about inheritance also need scope. A change in somatic cells belongs to those cells and their descendants; a change entering a reproductive lineage may have a different intergenerational reach. Neither a target observation nor an editing label establishes that reach, a clinical benefit or acceptable consequences. Molecular outcomes, organismal evidence and relevant governance remain connected questions with distinct information needs.

Worked example

Check your understanding

1. Does every genome-editing approach require cutting both DNA strands?

   Show reasoning

   No. DNA-cleavage and repair approaches are one class, while base-editing research demonstrates changes without requiring double-strand DNA cleavage. The distinction concerns mechanism and scope. It does not establish that any editing approach is universally precise, that every intended outcome occurs or that consequences have been evaluated.

2. Why are fourteen desired outcomes not fourteen outcomes without the assigned elsewhere change?

   Show reasoning

   Two of the fourteen also have that elsewhere attribute. The desired-only count is twelve, and its fraction is 12/20 rather than 14/20. The categories coexist in the supplied complete inventory, so a desired target result cannot be treated as a mutually exclusive alternative to every other change.

3. Why do seven tested negatives and twelve untested entries fail to establish nineteen negative outcomes?

   Show reasoning

   Only the seven tested negatives supply that result for the assigned elsewhere attribute. Twelve entries contain no relevant evidence and remain unknown. The tested identities are not stipulated representative either. Combining missing and negative information would both invent results and change the denominator of the stated comparison.

What to carry forward

Target intention, actual outcome, joint attributes, testing coverage and biological consequences must retain their own meanings.