PT  - JOURNAL ARTICLE
AU  - Sandeep Venkataram
AU  - Diamantis Sellis
AU  - Dmitri A. Petrov
TI  - Predictability of adaptive evolution under the successive fixation assumption
AID  - 10.1101/001016
DP  - 2015 Jan 01
TA  - bioRxiv
PG  - 001016
4099  - http://biorxiv.org/content/early/2015/07/28/001016.short
4100  - http://biorxiv.org/content/early/2015/07/28/001016.full
AB  - Several recent experimental studies assessed the likelihood of all possible evolutionary paths between ancestral and evolved sequences. All of these studies measured the fitness of the intermediate genotypes and assumed that the advantageous genotypes fix in the population before acquiring the next adaptive mutation along the path. Unfortunately, the successive fixation assumption used by these studies is typically invalid, given that natural selection often maintain alleles at intermediate frequency by a variety of mechanisms such as frequency-dependent selection, local adaptation, clonal interference, and fitness overdominance. Here we simulate adaptive walks using Fisher’s geometric model in diploid populations where previous work has shown that adaptation commonly generates balanced polymorphisms through overdominant mutations. We use these simulations to show that the use of the successive fixation assumption in this simple model is largely justified if the goal is to separate viable and inviable paths from each other. However, the estimates of the relative likelihoods of the viable paths become unreliable. We also show that the presence of balanced states along the true path significantly affects the number and likelihood distribution of viable paths when compared to walks without balanced states. These simple simulations highlight the importance of considering the effect of polymorphisms during adaptation especially given the prevalence of functional polymorphisms in natural populations.