Astronomy is a field uniquely afflicted by a limited ability to design and run controlled experiments. Hydrodynamical simulations of galaxy formation and evolution are used to alleviate this problem and are widely used in contemporary astronomy. They help interpret observations, validate methodologies, and make new predictions. This thesis concerns the use of simulations in addressing two outstanding questions at the frontier of the field. What are the systematic uncertainties influencing radio continuum weak gravitational lensing surveys such as those conducted with the upcoming Square Kilometer Array (SKA) telescope? What is the predicted dependence of galaxy clustering on various properties at fixed halo mass?\ud \ud I first measure the morphology and orientation of the radio continuum emitting component of central and satellite galaxies in the EAGLE simulation suite, a good proxy for which is the star-forming gas. Approximating the 3-dimensional morphology as an ellipsoid whose axis ratios are specified by the eigenvalues of the moment of inertia tensor, I show that the star-forming gas characteristically takes the form of a flattened disc at z=0. This is in contrast to the morphology of the stars, which is typically more extended along the minor axis. There exists, however, a significant correlation between the morphology of the two components in individual galaxies, with flatter stellar distributions generally being associated with flatter star-forming gas distributions. The difference in 3-dimensional morphology between the two components results in differences in their projected ellipticities, with the star-forming gas exhibiting a broader spread. The consequence of this for weak lensing surveys is that there will likely be a larger shape noise for galaxies as characterised by radio continuum emission than is the case for the optical. Assessing the redshift dependence, I show that the morphology of the star-forming gas comprising the progenitors of present day ~M* galaxies (i.e. those with stellar mass similar to that which defines the knee of the galaxy stellar mass function) shows significant evolution, with flattening increasing with time. I show that this is also the case in projection for all galaxies regardless of mass, which indicates that a redshift-dependent shape-fitting algorithm is likely necessary to measure the shapes of real galaxies in SKA-based surveys.\ud \ud Characterising orientation in terms of the minor axis direction, I present results showing that the star-forming gas is preferentially aligned with its host dark matter halo, however the degree of this alignment is found to be weaker than is the case between stars and dark matter. I provide fitting functions to the distribution of star-forming gas-dark matter alignment angles, which may be applied to semi-analytic models to more realistically model the intrinsic alignment (IA) effect in far larger volumes than can be followed by the current generation of state of the art hydrodynamical simulations. The morphological minor axis of the star-forming gas is found to align strongly with its kinematic axis, affording a route to observational identification of the unsheared morphological axis.\ud \ud The internal alignment between a galaxy and its host halo has significant implications for the IA of galaxy pairs, a key source of systematic bias in cosmic shear measurements. Using EAGLE, I measure the IA of galaxies as characterised by their star-forming gas. I find that in 3-dimensions and in projection the same qualitative results hold: the star-forming gas IA is weaker at all galaxy pair separations than is seen for the stars, however it is non-negligible when one considers orientation-direction alignment. IA strength as characterised by this measure increases with decreasing galaxy pair separation, following the same general trend as the stars. I find that the strong IAs seen at short pair separation are driven primarily by the one-halo term associated with central-satellite pairs in the haloes that host M* central galaxies. At fixed comoving separation, the radial alignment is stronger at higher redshift. My findings imply that the systematic uncertainty due to IA may be less severe in radio continuum weak lensing surveys than in optical counterparts, and that this stems from former’s tendency to be less well aligned with the dark matter structure of galaxies than the latter. Alignment models equating the orientation of star-forming gas discs to that of stellar discs or the DM structure of host subhaloes will therefore overestimate the impact of IAs on radio continuum cosmic shear measurements.\ud \ud I next address the topic of galaxy assembly bias. This is a consequence of the well-known prediction of halo assembly bias from simulations of the LCDM concordance model of cosmology, however it has yet to be conclusively observed in galaxy surveys. I first measure the galaxy assembly bias effect in EAGLE, and then assess the degree of correlation between various galaxy/halo properties with halo assembly time at fixed halo mass. Most properties which correlate with assembly time are found to be generally associated with galaxy clustering strength, however some properties which are not correlated with assembly time also exhibit secondary biases in galaxy clustering. This indicates that in EAGLE, not all of the secondary biases in galaxy clustering can be attributed to halo assembly time.\ud \ud While contemporary hydrodynamical simulations appear to agree in the existence of galaxy assembly bias and secondary bias, the exact nature of the signal varies, seemingly due to differences in the implementation of baryonic processes. I find that the EAGLE, BAHAMAS, and IllustrisTNG simulations agree in the prediction that galaxies with higher stellar mass cluster more strongly at fixed halo mass, EAGLE and IllustrisTNG predict a larger effect than is seen in BAHAMAS. The halo mass scale at which the secondary bias signal exhibits an inflection point is lower in EAGLE than is seen in IllustrisTNG. \ud \ud I next explore the possibility of using measures of environmental density to probe the dependence of clustering on properties at fixed halo mass. For such probes to be unbiased, it is required that measures of environmental density are uncorrelated with halo mass. I confirm this is the case with rho_N for M200