The Zen of Mutagenesis

"A selection is worth a thousand screens." David Botstein

What is the probability of finding a mutation in one particular gene on the chromosome?

Spontaneous mutations in a particular gene are typically very rare. Treatment with mutagens can increase the probability of getting a mutation in the gene, but it also increases the probability of getting other mutations in the chromosome as well.

Consider a typical gene in an average sized bacterial chromosome:

If (i) there is an equal, random chance of a "hit" (mutation) in any bp on the chromosome, and (ii) there is an average of one hit per cell, then the probability that a mutation will occur in any particular gene is:

However, many "hits" result in silent mutations -- for example, some mutations do not change the amino acid sequence of the resulting polypeptide, some mutations do not affect critical amino acid residues of the resulting polypeptide, and some mutations will be in unessential, noncoding sequences. About 90% of all base substitution mutations are typically silent mutations. Thus, the probability that a phenotypically detectable mutation will occur in any particular gene is:

Hence, if you screened 250 colonies per plate, it would take 200 plates to find a single mutant!

If you mutagenized the cells to increase the probability of finding mutations in a particular gene, the number of mutations elsewhere in the chromosome will increase proportionally. For example, heavier mutagenesis might increase the probability of finding a desired mutant 100-fold, so you would only have to look through 2 plates to find a each mutant. However, every mutant cell would have about 100 other mutations in addition to the desired mutation. Some of these other mutations may have insidious effects on the observed phenotype.

Take-home points:

So, why would anyone do heavy mutagenesis?

The need for mutagenesis depends upon whether a good selection is available for the desired mutant and how rare the desired mutatation is. For example, in contrast to the example above where a mutation anywhere in a gene was assumed to produce a mutant phenotype, certain phenotypes are only observed for mutations in a few specific bp within a gene. If the frequency of such rare mutations was 1% of that described in the example above, it would require 20,000 plates to find one mutant! To screen unmutagenized cells for such rare mutations would be very expensive and very boring (and not very smart).

Often a geneticist imagines a potential mutant phenotype, then designs a selection or screen to find the desired kind of mutant. If no one has isolated such a mutant before, you don't know if it will be possible to find such a mutant or how rare it will be until you try. It is useful to answer this question quickly to determine if the idea is worth pursuing. An initial experiment using heavily mutagenized cells is a practical way of quickly answering the questions "Is it possible to get such mutants?" and "How rare are such mutants?" Then the experiment can be effectively repeated with cells that have not been as heavily mutagenized. (Alternatively, the first experiment may tell you that the idea was flawed and you can think up some other scheme for a mutant hunt.)

Return to Microbial Genetics supplement.

Please send comments, suggestions, or questions to
Last modified July 2, 2004