Computational protocol: Contrasting Epidemic Histories Reveal Pathogen-Mediated Balancing Selection on Class II MHC Diversity in a Wild Songbird

Similar protocols

Protocol publication

[…] All microsatellite markers used had a null allele frequency less than 0.10 in our population. Two out of 32 population-locus combinations significantly deviated from the assumptions of Hardy-Weinberg equilibrium (α = 0.002; Hofi20 in the post-epidemic eastern population and Hofi52 in the pre-control western population). Two of twenty-eight locus pairs showed significant linkage disequilibrium (α = 0.002; Hofi3 and Hofi5; Hofi20 and Lox3). However, population-level comparisons revealed that LD between these locus pairs was only detected in one of four populations (pre-epidemic eastern), suggesting that the loci are unlikely to be tightly linked. Microsatellite diversity indices were analyzed using GENEPOP v. 3.4 and FSTAT v2.9.3.2 . [...] We used JMP 9.0 (SAS Institute, Cary, NC) for all non-molecular statistical analyses. For the experimental challenge data, we used mixed linear models to examine associations between individual-level variant diversity and disease severity across individuals in our experimental population. We included bird identity as a random effect and microsatellite heterozygosity, which has been shown to influence disease severity in house finches , as a covariate. Because prior studies have found that intermediate levels of MHC diversity are associated with the highest pathogen resistance , , we included both first and second-order relationships between individual-level MHC diversity and MG resistance in our model. We used a log-likelihood ratio test evaluated against a Chi-square distribution to determine whether including a second-order relationship for MHC diversity improved model fit.To examine temporal changes in individual-level MHC diversity and microsatellite diversity (allelic diversity and heterozygosity) prior to and following the epidemic, we used general linear models, including population (western versus eastern), time (pre- versus post-epidemic), and their interaction in our model. For the microsatellite analyses, we considered locus-specific metrics of allelic diversity and heterozygosity for each population as independent data points in our model.We defined rare MHC variants as those that were only present in a single individual in each population (e.g. pre-epidemic east, post-epidemic east, pre-control west, post-control west). To compare the proportion of rare MHC variants in each population, we standardized for the differences in sample size among populations () by repeatedly subsampling our eastern population data sets, which were significantly larger. We used a random number generator to subsample without replacement, generating a sample size of n = 8 for the pre-epidemic eastern population and a sample size of n = 14 for our post-epidemic eastern population. The number of rare variants in each of 10 eastern subsamples were averaged and compared to the true values generated from the equivalently sized western populations.We distinguished the 30 codons that likely make up the peptide-binding region (PBR) , which is expected to be under strong positive selection due to its role in antigen binding specificity . We conducted separate molecular sequence analyses on the PBR and non-PBR regions using MEGA 4.0 and the Jukes-Cantor model of sequence evolution . For all analyses, we used 1,000 permutations where necessary and assumed the Kimura 2P model of evolution.We used Arlequin 3.1 to conduct analyses of molecular variance , and to quantify pairwise Fst values across our four populations (pre-epidemic eastern, post-epidemic eastern, pre-control western, post-control western) via Mantel tests. For the genetic structure analyses, we predicted that the MG epidemic would result in significant genetic structure between pre- and post-epidemic eastern populations whereas we would not detect genetic structure between control populations sampled over the same time period (pre-control versus post-control western). […]

Pipeline specifications

Software tools Genepop, MEGA, Arlequin
Application Population genetic analysis
Organisms Haemorhous mexicanus
Diseases Conjunctivitis, Encephalitis, Arbovirus, Mycoplasma Infections