Computationally unraveling genetic factors underlying morphological patterns of wheat’s root growth: a scientific data analysis approach

Abstract A collection of 53 wheat varieties’ root-growth hourly trajectories is clustered into seven distinct morphological root growth patterns. From more than 30K DNA variants of these wheat varieties, pair-wise comparisons of root growth pattern are carried out to demonstrate varying signal-to-noise ratios. Each short serial genotypic combination of various lengths as a piece of classifying information is computed and confirmed by passing a reliability check. Shannon entropy-based computational paradigm called Categorical Exploratory Data Analysis (CEDA) is employed to accommodate the entire categorical datatype. The reliability check is devised based on two ensembles of mimicking observed and simulating null contingency tables to respectively give rise to the alternative and null distributions of entropy with their overlapping area being equal to the minimum sum of Type-I and Type-II errors. Then, a bipartite network between wheat varieties and selected genotypic combinations is constructed as a heatmap of presence-absence memberships to further reveal bipartite interacting relations. As such another contrasting heatmap for members outside of the targeted branch-pair displays outlier classification information. Upon all selected genotypic combinations from the 21 pairs of branch-vs-branch classifications, two genes near identified loci are identified with existing literature for biological relevance.