Theoretical Biology & Bioinformatics

tree of life
Rob J. de Boer

Rob J. de Boer

Home | Research | Publications | Books | CV |

Estimating relative fitness from competition experiments

Marée et al (J. Virol. 2000, 74: 11067-11072) develop a formula for estimating the relative fitness (1+s) from virological competition experiments (Download PDF). Lets call one variant the wild type and the other the mutant. Virus need not be growing exponentially during the experiment. The formula uses the fold expansion of the wild type virus, and the fold change in the ratio of the genotype frequencies. This is all computed for you if you provide the total virus concentrations, and the percentages of mutant virus, at the start and at the end of the experiment. Because we require fold changes only, the virus concentration can be provided in arbitrary units (like P24, RNA, DNA, etcetera), as long as your assay is linear in the viral density.

In case you refresh during the experiment, i.e. dilute the productively infected cells, provide the dilution factor. One means no dilution, two means a 2-fold dilution (i.e., a continuation of the competition experiment with half of the cells), etcetera.

If you have an estimate of the death rate (delta) of productively infected cells, you should definitely provide this. In case you have no clue, you may enter a zero. This boils down to assuming that there is a selection pressure on the net replication rate of the two variants in the experiment.

We provide two variants of the algorithm. The first assumes that you did a quantative assay providing the percentages of mutant and wild type virus. The second assumes that you sampled a number of sequences, and that you know the number of mutant and wild type sequences. Importantly, since sampling sequences is typically noise, the latter version calculates confidence limits. So either click Percentages or n Samples to continue.

Both versions of the algorithm allow for defining a "restgroup" of irrelevant mutants.

Important improvement

Bonhoeffer et al (Proc. R. Soc. Lond. B. Biol. Sci., 2002: 269, 1887-1893) have extended this method to allow for time series data (Download PDF). This new method is also publically available.