Protein Networks as Logic Functions in Development and Cancer

Abstract

Many biological and clinical outcomes are based not on single proteins, but on modules of proteins embedded in protein networks. A fundamental question is how the proteins within each module contribute to the overall module activity. Here, we study the modules underlying three representative biological programs related to tissue development, breast cancer metastasis, or progression of brain cancer, respectively. For each case we apply a new method, called Network-Guided Forests, to identify predictive modules together with logic functions which tie the activity of each module to the activity of its component genes. The resulting modules implement a diverse repertoire of decision logic which cannot be captured using the simple approximations suggested in previous work such as gene summation or subtraction. We show that in cancer, certain combinations of oncogenes and tumor suppressors exert competing forces on the system, suggesting that medical genetics should move beyond cataloguing individual cancer genes to cataloguing their combinatorial logic.

Revealing Shared and Distinct Gene Network Organization in Arabidopsis Immune Responses by Integrative Analysis

“…NGF assigns a score called IS to each gene/ interaction according to its contribution to improve the classification accuracy. The complete methodology is detailed in the work of Dutkowski and Ideker (2011). The gene expression profiles of PTI, the control, and AraONE were used to train a PTIrelated NGF model, whereas the gene expression profiles of ETI, the control, and AraONE were used to train an ETI-related NGF model.…”

“…State-of-the-art machine-learning methods have been used to successfully identify stress-responsive genes (Ma et al, 2014;Shaik and Ramakrishna, 2014) and development-related gene associations (Bassel et al, 2011a) from large-scale plant gene expression data. Recently, Dutkowski and Ideker (2011) proposed a new machine-learning ranking algorithm called the networkguided forest (NGF). NGF was developed from Random Forest (RF; Breiman, 2001), a popular machine-learning algorithm that uses many slightly different decision trees to infer the relationship between input features and class labels.…”

“…In the NGF, the molecular network is further introduced into RF to guide the growth of each decision tree. For example, by jointly using gene expression states and molecular network information as input to build the classification model, NGF can identify more cancer susceptibility genes in cancer stratification compared with conventional methods that only consider gene expression differences (Dutkowski and Ideker, 2011).…”

Edge biomarkers for classification and prediction of phenotypes

“…In this method, the modules are extracted from the protein interaction network by a Markov clustering algorithm (MCL), and their activities are defined as the degree-weighted mean expression of genes in each module. Dutkowski et al used the network-guided forests (NGF) algorithm to identify logic functions (e.g., decision trees) with the same topological structure as the sub-networks in a prior network [45]. NGF is expected to connect the activity of each predictive module to the activity of its component genes.…”

Discovery and characterisation of dietary patterns in two Nordic countries

“…56,57 Because of its popularity in many areas special and/or enhanced varieties of RF have appeared. [58][59][60] The second of the two families of predictive modelling, brought into play in the study outlined here, is known as Nearest Shrunken Centroid (NSC) classifier. 55 It is widely appreciated due to its relatively low computational complexity during design/learning and execution as well as its inherent ability to perform variable subset selection, as part of the design/learning procedure.…”