Abstract:
|
Immune cells in tumor microenvironment have important effect on tumor progression, thus can serve as attractive prognostic or therapeutic targets. The identification of therapeutic pathways in tumor and infiltrating immune cells that associate with positive outcomes, such as survival, is a major challenge. Classical pathway analysis relies on pre-defined set of pathways, which makes novel discovery challenging. In this work, we relax this assumption of the pre-defined pathway, and explore an approach that leverages the fully connected KEGG database as a single entity. Tissue-specific weights are appended to the network. The weighted network is partitioned using an attribute-based module detection algorithm, which are used routinely with PPI networks. Analysis of this type captures fundamentally different aspects of the enrichment signal. Since it does not rely on pre-defined categories (pathways) for individual testing, there is an opportunity to expose novel pathways and crosstalk. We leverage this approach for the elucidation of tissue-specific pathways that associate with survival outcomes and compare our findings to classical topological approaches to enrichment testing.
|