Research methods that predict Fe bioavailability for humans can be extremely useful in evaluating food fortification strategies, developing Fe-biofortified enhanced staple food crops and assessing the Fe bioavailability of meal plans that include such crops. In this review, research from four recent poultry (Gallus gallus) feeding trials coupled with in vitro analyses of Fe-biofortified crops will be compared to the parallel human efficacy studies which used the same varieties and harvests of the Fe-biofortified crops. Similar to the human studies, these trials were aimed to assess the potential effects of regular consumption of these enhanced staple crops on maintenance or improvement of iron status. The results demonstrate a strong agreement between the in vitro/in vivo screening approach and the parallel human studies. These observations therefore indicate that the in vitro/Caco-2 cell and Gallus gallus models can be integral tools to develop varieties of staple food crops and predict their effect on iron status in humans. The cost-effectiveness of this approach also means that it can be used to monitor the nutritional stability of the Fe-biofortified crop once a variety has released and integrated into the food system. These screening tools therefore represent a significant advancement to the field for crop development and can be applied to ensure the sustainability of the biofortification approach.

Synonymous variants within coding regions may influence protein expression and function. We have previously reported increased protein expression levels ex vivo (~120% in comparison to wild-type) from a synonymous polymorphism variant, c.354G>A [p.P118P], of the ADAMTS13 gene, encoding a plasma protease responsible for von Willebrand Factor (VWF) degradation. In the current study, we investigated the potential mechanism(s) behind the increased protein expression levels from this variant and its effect on ADAMTS13 physico-chemical properties. Cell-free assays showed enhanced translation of the c.354G>A variant and the analysis of codon usage characteristics suggested that introduction of the frequently used codon/codon pair(s) may have been potentially responsible for this effect. Limited proteolysis, however, showed no substantial influence of altered translation on protein conformation. Analysis of post-translational modifications also showed no notable differences but identified three previously unreported glycosylation markers. Despite these similarities, p.P118P variant unexpectedly showed higher specific activity. Structural analysis using modeled interactions indicated that subtle conformational changes arising from altered translation kinetics could affect interactions between an exosite of ADAMTS13 and VWF resulting in altered specific activity. This report highlights how a single synonymous nucleotide variation can impact cellular expression and specific activity in the absence of measurable impact on protein structure.

A new empirical Bayes approach to variable selection in the context of generalized linear models is developed. The proposed algorithm scales to situations in which the number of putative explanatory variables is very large, possibly much larger than the number of responses. The coefficients in the linear predictor are modeled as a three-component mixture allowing the explanatory variables to have a random positive effect on the response, a random negative effect, or no effect. A key assumption is that only a small (but unknown) fraction of the candidate variables have a non-zero effect. This assumption, in addition to treating the coefficients as random effects facilitates an approach that is computationally efficient. In particular, the number of parameters that have to be estimated is small, and remains constant regardless of the number of explanatory variables. The model parameters are estimated using a Generalized Alternating Maximization algorithm which is scalable, and leads to significantly faster convergence compared with simulation-based fully Bayesian methods.

One of the most common problems facing empirical researchers is when a portion of the data is missing. We will review three different types of ‘missingness’, namely missing completely at random, missing at random and missing not at random, and we will discuss how missing data can affect data analysis. We review methods to deal with missing data, including the simple ‘complete-case analysis’ approach, in which we only use the observations in the data set for which all the data is available, and the more sophisticated ‘multiple imputation’ approach, in which we repeat the analysis using multiple (completed) copies of the data set, and obtain the estimates of interest by averaging across all analyses. We will demonstrate how to implement solutions to missing data and review the limitations of the methods.