Disease progression and solid tumor survival: a transcriptome decoherence model

Abstract

Networks of genes are typically generated from expression changes observed between control and test conditions. Nevertheless, within a single control state many genes show expression variance across biological replicates. These transcripts, typically termed unstable, are usually excluded from analyses because their behavior cannot be reconciled with biological constraints. Grouped as pairs of covariant genes they can however show a consistent response to the progression of a disease. We present a model of coherence arising from sets of covariant genes that was developed in-vitro then tested against a range of solid tumors. DGPMs, Decoherence Gene Pair Models, showed changes in network topology reflective of the metastatic transition. Across a range of solid tumor studies the model generalizes to reveal a richly connected topology of networks in healthy tissues that becomes sparser as the disease progresses reaching a minimum size in the advanced tumors with minim survivability.