Protein-protein interaction (PPI) networks are being increasingly used to support functional genomic research. PPI networks can consist of several thousand nodes and sampling is often used to extract meaningful information representative of the global network. However there has been relatively little research carried out on the impact of sampling and significance of depth on such networks. In this study, six PPI networks, three relevant to heart failure, one to asthma, and two consisting of randomly-selected proteins, are analyzed and compared through different network levels. The effect of network depth is examined in terms of network metrics, i.e. degree and betweenness centrality, and on the classification methods for identifying potentially significant nodes, which may represent novel therapeutic targets.