Spatial Analysis of Functional Enrichment (SAFE) for the Saccharomyces cerevisiae Genetic Interaction Similarity Network - Graph Layout Effects

Abstract

Spatial Analysis of Functional Enrichment (SAFE) is a computational technique that wasdeveloped to automatically discern and label functional structure is biological networks [5, 3, p. 2]. SAFE is the first method of its kind. Biological networks are notoriously complex and voluminous data sets and some have characterized them as analytically "impenetrable" to biologists [3, p. 3]. For example, researchers have recorded extensive genetic interaction networks for the budding yeast, Saccharomyces cerevisiae;These networks map physical, biochemical and phenotypic relationships between nearly all genes in the Saccharomyces cerevisiae genome [9, 8, 12, 14–16, 10, 3, p. 3]. Random effects due to layout techniques potentially compromise the reliability of SAFE’s functional annotations. This study aims to quantify the extent to which stochastic effects the reliability of SAFE analysis. Further, this study should justify further study of stochastic effects. This study aims to move towards a common framework for evaluating layouts for genetic interaction similarity networks. For three stochastic layout techniques 10 SAFE analyses were performed. For the Kamada-Kawai (KK), Cytoscape Prefuse Force-Directed (PFD) and Cytoscape Circular (CC) layout, the result SAFE functional annotations were analyzed for randomness using Spearman’s Rank Correlation. While the PFD may be more cumbersome to visually evaluate thanKK, with PFD, more functional annotation terms are exhibit consistency across randomlayouts. This work can be easily expanded in several ways. One obvious extension of thiswork is to carry the Spearman’s Rank Correlation test down through more detailed levels of resolution in the budding yeast GIS network. Exploring deterministic layout algorithms and layout-independent neighborhood metrics is also of critical importance.