Back

Incorporating selfing to purge deleterious alleles in a cassava genomic selection program

Somo, M.; Jannink, J.-L.

2020-04-05 genetics
10.1101/2020.04.04.025841 bioRxiv
Show abstract

Cassava has been found to carry high levels of recessive deleterious mutations and it is known to suffer from inbreeding depression. Breeders therefore consider specific approaches to decrease cassavas genetic load. Using self fertilization to unmask deleterious recessive alleles and therefore accelerate their purging is one possibility. Before implementation of this approach we sought to understand better its consequences through simulation. Founder populations with high directional dominance were simulated using a natural selection forward simulator. The founder population was then subjected to five generations of genomic selection in schemes that did or did not include a generation of phenotypic selection on selfed progeny. We found that genomic selection was less effective under the directional dominance model than under the additive models that have commonly been used in simulations. While selection did increase favorable allele frequencies, increased inbreeding during selection caused decreased gain in genotypic values under the directional dominance. While purging selection on selfed individuals was effective in the first breeding cycle, it was not effective in later cycles, an effect we attributed to the fact that the generation of selfing decreased the relatedness of the genomic prediction training population from selection candidates. That decreased relatedness caused genomic prediction accuracy to be lower in schemes incorporating selfing. We found that selection on individuals partially inbred by one generation of selfing did increase mean genetic value of the partially inbred population, but that this gain was accompanied by a relatively small increase in favorable allele frequencies such that improvement in the outbred population was lower than might have been intuited.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.