Digitizing omics profiles by divergence from a baseline

• DOI: 10.1073/pnas.1721628115
• Publication Date: 2018-04-17T15:00:35Z
• Authors: Wikum Dinalankara; Qian Ke; Yiran Xu; Lanlan Ji; Nicole Pagane; Anching Lien; Tejasvi Matam; Elana J. Fertig; Nathan D. Price; Laurent Younes; Luigi Marchionni; Donald Geman
• Keywords: Omics; Baseline (sea); Divergence (linguistics)

Data collected from omics technologies have revealed pervasive heterogeneity and stochasticity of molecular states within between phenotypes. A prominent example such occurs genome-wide mRNA, microRNA, methylation profiles one individual tumor to another, even a cancer subtype. However, current methods in bioinformatics, as detecting differentially expressed genes or CpG sites, are population-based therefore do not effectively model intersample diversity. Here we introduce unified theory quantify sample-level that is applicable single profile. Specifically, simplify an profile digital representation based on the set samples reference baseline population (e.g., normal tissues). The state any subprofile expression vector for subset genes) said be "divergent" if it lies outside estimated support distribution consequently interpreted "dysregulated" relative baseline. We focus two cases: features distinguished subsets regulatory pathways). Notably, since divergence analysis at sample level, dysregulation can analyzed probabilistically; example, estimate probability gene pathway divergent some population. Finally, reduction complexity facilitates more "personalized" biologically interpretable variation, illustrated by experiments involving tissue characterization, disease detection progression, disease-pathway associations.