How many samples do you need to investigate relationships between genetic make-up & immune function?

When an organism is exposed to a pathogen, what determines their ability to resist or recover from infection? Mounting an effective immune response is a complex dance with multiple partners, changing tempos, and maybe even a costume change or two: An adaptive immune response results from the interplay of the surrounding environment and the specific individual, that individual’s (and sometimes the parents!) history of infection, and the genetic make-up of the individual, just to name a few of the partners. Scientists have been investigating the effects of immunogenetic diversity and genotype on immune system functionality for decades. Among the most studied immune system genes are those of the Major Histocompatibility Complex (MHC), which encode the cell-surface proteins responsible for distinguishing between ‘self’ and ‘non-self’ peptides (Piertney & Oliver 2006). Thus, MHC genes control, in part, an organism’s ability to respond to pathogens through the adaptive immune system. I did my Ph.D. on MHC genes, so I still have a soft spot in my heart for this set of genes; Also, I really love a paper that makes tasty lemonade out of lemons, as a recent paper by Arnaud Gaigher and colleagues (2019) did.

First Author Arnaud Gaigher, Photo credit:

The authors explored MHC-mediated immunocompetence in a wild population of Barn owls (Tyto alba), taking the very cool micro-level approach of measuring response to an immune challenge as opposed to the more traditional macro-level approach of measuring overall fitness proxies like growth, survival, or reproduction. The authors used data from a series of immune challenges conducted on barn owl nestlings from a population that have been regularly captured, ringed, and sampled for blood since 1996 (Roulin et al. 2007). They measured immunocompetence using several methods: 1) one cohort of nestlings were vaccinated with sheep red blood cells (SRBCs), human serum albumin (HSA), and tetanus toxoid (TT), which are all nonpathogenic antigens expected to activate the humoral immune pathway (Janeway et al. 2001) and then the concentration of antibodies, haematocrit, and leucocytes in their blood was measured repeatedly over 18 days. 2) A second cohort of nestlings underwent the PHA (phytohaemagglutinin: a plant lectin) skin test, which involves injection of PHA under the skin to induce a cutaneous inflammatory reaction and then swelling is measured over time. Immunoglobulin protein concentrations were also measured in this cohort. The authors assessed MHC diversity via next-generation sequencing of several MHC genes (MHC-I and MHC-IIB DAB1 and DAB2) targeting specifically the exons that contain the peptide binding region responsible for binding different pathogen peptides.

A barn owl – Photo by Matt Davis –

Like many studies examining MHC diversity, the authors found a large amount of MHC variation among the >450 barn owl nestlings: 69, 25 and 17 alleles for MHC-I, MHC-IIB DAB1 and DAB2 respectively, >60% of the population possessed four MHC-I alleles, and >70% of the population were heterozygous for each MHC-II gene. Because 1) the genetic code is degenerate, meaning different DNA sequences can code for the same amino acid and 2) some amino acids have similar functional properties, meaning that one amino acid sequence might bind very similar peptides, MHC alleles are generally collapsed into ‘supertypes’, which are groups of alleles with similar peptide-binding properties.

The authors then compared MHC diversity and the possession of the most frequent MHC supertypes with the response to each immunocompetence challenge. After adjusting for multiple comparisons, neither overall MHC diversity nor the possession of the nine most frequent MHC supertypes predicted variation in the immunocompetence parameters measured. The average effect size for MHC diversity predictors in their best statistical models was 0.09, which is quite low (small effect sizes are generally considered ~0.1), and 0.07 for MHC supertypes. The authors then used power analyses and simulated dependent variables to estimate the power their sample size had to detect different effect sizes. While their tests had the power to detect moderate to large effect sizes (0.3-0.8), they had little power to detect low effect sizes: The power to detect their observed effect sizes was only 0.16 and 0.17 respectively. Thus, moderate-to-strong associations between MHC and immunocompetence were unlikely in their study, but that they could not rule out small associations.

Figure 1A replicated from Gaigher et al. 2019 shows the power of the authors’ study to detect associations between MHC diversity and immunocompetence at different effect sizes, with the conventional power threshold of 0.8 shown as the dashed line.

This finding has serious implications for past and future work because their sample size is as large or larger than previously published studies examining the MHC-immunocompetence link. The authors then turned these ‘lemon’ results into a really cool recommendation about the statistical power needed to find the connections between genetic make-up and immune function in wild populations. They examined the power to detect a relationship between MHC and immunocompetence in a range of studies in wild populations, with samples sizes from 10 to >200. Most studies had the power to detect large to moderate effects, but not small, whereas most studies reported an average effect size of 0.145.

The power to detect an association also depends on the variation in MHC diversity and the frequency of MHC alleles in a population, so the authors then modeled the sample size needed to detect different effect sizes, for studies using MHC diversity and also the possession of specific alleles (Figure 3). For moderate to large associations between immunocompetence and MHC diversity, ~ 80 samples are needed. However, for small effect sizes of MHC diversity, nearly 800 samples are needed. If researchers are interested in specific alleles, we may need as many as 4,000 samples to detect small effects if an allele is rare. This study provides new insights for research investigating host–pathogen interactions and highlights the lack of statistical power as a major limitation to understand how MHC might be affecting immunocompetence in vertebrates. Moreover, this study provides a concrete roadmap for estimating sample sizes needed in future work.

Figure 3 replicated from Gaigher et al. 2019


Janeway, C. A., Travers, P., Walport, M., & Shlomchik, M. J. (2001). Immunobiology: The immune system in health and disease. New York, NY: Garland Science.

Piertney, S. B., & Oliver, M. K. (2006). The evolutionary ecology of the major histocompatibility complex. Heredity, 96, 7–21. https ://

Roulin, A., Christe, P., Dijkstra, C., Ducrest, A. L., & Jungi, T. W. (2007). Origin-related, environmental, sex, and age determinants of immunocompetence, susceptibility to ectoparasites, and disease symptoms in the barn owl. Biological Journal of the Linnean Society, 90, 703–718. https ://

This entry was posted in birds, next generation sequencing and tagged , , . Bookmark the permalink.