Activation of telomere maintenance mechanisms (TMMs) is a crucial factor for indefinite proliferation of cancer cells. The most common TMM is based on the action of telomerase, but in some cancers telomeres are elongated via homologous recombination based alternative mechanism (ALT). Despite their importance, little is known about TMM regulation and factors responsible for TMM phenotype choice in different cells. Currently, many studies address the involvement of few genes in TMMs, but a consensus unified picture of the full process is missing.We have developed a computational biology framework combining knowledge-and data-driven approaches to aid in understanding of TMMs. It is based on a greedy algorithm with three core modules: (1) knowledge-based construction/modification of molecular pathways for telomerase-dependent and alternative TMMs, (2) coupled with gene expression data-based validation with an in-house pathway signal flow (PSF) algorithm, and (3) iteration of these two coupled steps until converging at pathway topologies that best reflect state of the art knowledge and are in maximum accordance with the data. We have used gene expression data derived from cell lines and tumor tissues and have performed extensive literature search and multiple cycles of greedy iterations until reaching TMM assessment accuracy of 100% and 77%, respectively.Availability of TMM pathways that best reflect recent knowledge and data will facilitate better understanding of TMM processes. As novel experimental findings in TMM biology emerge, and new datasets are generated, our approach may be used to further expand/improve the pathways, possibly allowing for making distinctions not only between telomerase-dependent and ALT TMMs, but also among their different subtypes. Moreover, this method may be used