This dichotomy may be explained by recognition of the fact that our correlative and Granger analyses were pairwise, while feature selection methods were multivariate. Whereas such pairwise analyses have sensitivity for strong biological relationships that are relatively direct and linear, Nodakenin multivariate analyses may provide sensitivity towards complex and perhaps contingent relationships between regulatory features. Thus, it is conceivable that cell area may modulate the causal impacts of several other features upstream of cell speed, in such a way that its own causal influence is distributed and not directly discernable. It will therefore be crucial in the future to develop multivariate statistical implementations of Granger causality analysis for this setting, including improving sensitivity to non-linear causal interactions. Even more surprising was that both CMAC localization and cell shape features were regulated by cell speed. In general, these findings highlight ongoing questions regarding the possible orientations of causation relative to complexity in such complex biological systems, and indeed about the relative positions of such features in a hierarchy of complexity. Specifically, we were surprised in the first instance to find that faster cell Ginsenoside-Ro migration caused increased distances between CMACs and the cell edge, reducing the peripheral skew in CMAC localization. This was unexpected because our empirical findings refute previous perturbation-derived inferences that CMAC centralization may cause the inhibition of cell migration. Indeed, when combined with evidence that CMAC lifetime causally regulates cell speed, this finding allows the disentanglement of previously convoluted correlations between CMAC centralization, CMAC stabilization and cell migration. Thus, we find that CMAC stability strongly influences cell speed, which in turn modulates CMAC localization. Equally unexpected was the finding that cell speed lies upstream of cell shape determination, with faster migration in control cells promoting a less round cell shape. While the existence of a correlative relationship between cell shape and cell speed has been abundantly clear, the direction of causation has not. However, evidences that symmetry breaking and associated cell deformations necessarily precede cell migration promote the assumption that cell shape would act upstream of cell speed, rather than downstream as found here under all three experimental conditions. A further investigation of the relative magnitude and frequency of shape changes during symmetry breaking and at different migration speeds may more fully explain this surprising finding. In general, this observation reveals a possible limitation in the sensitivity of our causal analysis strategy, since each experimental conditions is described by a single causal wiring pattern. This means that potential variations in causal wiring arising within conditions may be lost. It is conceivable, for example, that wiring patterns may differ during alternate phases of migration and/or between different migratory modes. Indeed, the latter is hinted at by wiring changes induced following Rho-activation, as discussed further below. Methodologically, the ability to disaggregate data from within experimental conditions to identify recurring migratory phases/ modes may help to contextualize our understanding of causal network structures during evolving migratory behaviors, as previously exemplified during cell polarization.