Malaria is a major public health concern for the one-third of the human population esti- mated to be exposed to the threat of the most virulent species, Plasmodium falciparum. Modern molecular and computational tools from population genetics may help to better understand and fight the burden of drug resistant malaria. Malaria biology is substantially different from the underlying paradigm of standard population genetics models, most notably because malaria has both a asexual haploid phase and a sexual diploid phase where selfing (i.e. mating between genetically identical parasites) is possible. It is therefore fundamental to understand if commonly used population genetics methods are robust to the deviations from standard expectations imposed by the malaria life-cycle. We build novel models of malaria population genetics and provide guidelines to interpret empirical studies of the spread of drug resistance. Using realistic models of epistasis between genes involved in drug resistance we suggest that all signals of linkage disequilibrium (LD) are possible and that researchers should be confident in reporting a lack of statistical association between genes involved in resistance to antimalarials. We also suggest that researchers should be cautious in interpreting changes in the prevalence of drug resistance after control interventions as reductions in transmission can cause a change in prevalence without concomitant change in frequency of resistance. We provide guidelines to better design and interpret studies related to estimating effective population size (Ne). We use computational simulations to study scenarios that can approximate control and elimination interventions (i.e. where a significant part of the population is killed). For Ne estimation our results suggest that researchers must increase the number of individuals and loci genotyped in order to have sufficiently precise Ne estimates. LD-based Ne estimators are more appropriate for early detection of control and elimination interventions than temporal-based Ne estimators. Long- term estimators based on heterozygosity should not be used to make inferences about contemporary demographic processes. We also applied our analysis to disease vectors and we concluded that LD-based estimation is able to detect demographic seasonality patterns (i.e. changes in population size due to variations imposed by wet and dry seasons) whereas temporal estimators will provide averages over longer periods of time. We also studied selection detection using FsT-outlier approaches. Our results suggest that temporal FST might not be appropriate for early detection of genes involved in drug resistance (e.g. in the case of artesunate derivatives). We also provide software and guidelines to better design and interpret studies (also across other taxa) of selection based on FsT-outlier approaches. Most notably our results suggest that sampling only one or two SNPs per locus (as it is done in many empirical studies) might not be sufficient to detect areas of the genome under selection, and that at least 4 SNPs per loci should be genotyped.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:569901 |
| Date | January 2011 |
| Creators | Antão, Tiago Rodrigues |
| Publisher | University of Liverpool |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0025 seconds