Effects of Phenotypical Patterns on Epigenetic Markers

摘要

The Eastern Fence Lizards (specie S_1) are exposed to Fire Ants (specie S_2) in some areas of their habitat (space B_1) and not exposed in some other areas (space B_2.) The population of S_1 in B_1 have responded phenotypically in patterns that are both evolutionary (cross-generational) and plastic (within lifetime.) Existence of relationships, if any, between phenotypical patterns and epigenetic markers is the proof that patterns affect epigenetic markers and not DNA sequence. We investigate such relationships on a dataset (D) with 189 epigenetic markers and four phenotypes of Sex (SEX), Right-hind Limb (RHL), Snout Vent Length (SVL), and ratio of RHL/SVL collected for two groups (g_0 and g_1) of specie S_1 from B_1 and B_2, respectively. The goal is threefold: (a) whether there is a subset of epigenetic markers in D that differentiates between members in g_0 and g_1, (b) which subset is the best, if more than one such subset exists, and (c) whether there are epigenetic markers that significantly differ between the lizards in g_0 and g_1. Part (a) was met by introducing eight algorithms that identified eight subsets of epigenetic markers from which four strict and four relaxed representatives of D were generated. Part (b) was met by use of inductive learning algorithm C4.5. One of the eight algorithms (Entropy-Thinning) delivered the best representative (R) of D (with 14 markers.) R predicted four phenotypes separately with high accuracies (≥85%) as a proof of strong relationships between phenotypical patterns and markers. Part (c) was met by using One-Way Classification approach on R. Four epigenetic markers of Loci 036, 060, 071, and 101 were significantly differ (99.5% certainty) between g_0 and g_1.