Employing network science to understand particle interactions helps manufacture advanced materials with superior force transmission and heat transfer. However, knowledge of the dependence of networks on particle features such as shape is missing. This study computes particle shape ─ the average of three-dimensional sphericity and roundness, and multiscale network variables ─ degree, edge betweenness centrality and global clustering coefficient from unweighted/weighted contact and thermal networks ─ for three sands based on their X-ray computed tomography images. The dependence of network features on particle shape is explored for both individual particles and bulk sand samples. Results show that particle shape affects the degree in a network at sample and particle scales differently. In contrast, weighted edge betweenness centrality has a consistent inverse relationship with particle shape at both scales. The weighted edge betweenness centrality values from different samples consistently indicate that 20% of network edges (e.g., contacts) are responsible for 60% of the heat transfer in dry sands. Although unweighted edge betweenness centrality cannot reflect the heat transfer directly, it has a similar correlation with particle shape to the weighted feature. Global clustering coefficient from the thermal network increases in round particle packings and can indicate the mechanical rigidity of sands.