Uncovering biomarker genes with enriched classification potential from Hallmark gene sets

Abstract

Given the complex relationship between gene expression and phenotypic outcomes, computationally efficient approaches are needed to sift through large high-dimensional datasets in order to identify biologically relevant biomarkers. In this report, we describe a method of identifying the most salient biomarker genes in a dataset, which we call “candidate genes”, by evaluating the ability of gene combinations to classify samples from a dataset, which we call “classification potential”. Our algorithm, Gene Oracle, uses a neural network to test user defined gene sets for polygenic classification potential and then uses a combinatorial approach to further decompose selected gene sets into candidate and non-candidate biomarker genes. We tested this algorithm on curated gene sets from the Molecular Signatures Database (MSigDB) quantified in RNAseq gene expression matrices obtained from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) data repositories. First, we identified which MSigDB Hallmark subsets have significant classification potential for both the TCGA and GTEx datasets. Then, we identified the most discriminatory candidate biomarker genes in each Hallmark gene set and provide evidence that the improved biomarker potential of these genes may be due to reduced functional complexity.

Transfer transcriptomic signatures for infectious diseases

“…Our work and that of others also support the possibility of extending prediction across species; for example, a recent study indicates that the mouse transcriptome reveals potential signatures of protection and pathogenesis in human tuberculosis ( 38 ). The approach also found significant information content in more general sets of signatures, such as those listed in the hallmark gene sets of the Molecular Signatures Database, a result consistent with recent observations ( 39 ) and the use of co-expression patterns to define transcriptome modules ( 40 ). We interpret this as indication that even in the absence of pre-existing information in the literature, well-defined sets of hallmark genes will enable the extraction and creation of transfer signatures.…”

Diagnostic blood RNA profiles for human acute spinal cord injury

“…Despite the attractiveness of single-gene biomarkers, there is growing evidence (mainly from cancer research) to suggest that their performance is limited ( Targonski et al, 2019 ). Of course, genes do not work in isolation, and there is overwhelming evidence for reproducible transcriptional covariation in blood and other tissues, suggesting a modular organization to genomic function.…”

Exploration into biomarker potential of region-specific brain gene co-expression networks

“…We used a two phase, bottom-up classification approach of a feedforward neural network, known as Gene Oracle 15 ( https://github.com/SystemsGenetics/gene-oracle ), to classify brain regions, and thus, identify the region-specific gene biomarkers. Gene Oracle uses a multilayer perceptron (MLP) feedforward neural network 35 to identify biomarker gene sets with a significant classification accuracy when comparing to sets with equal number of random genes.…”

“…The effectiveness of the biomarkers to discriminate conditions can be formally tested using machine learning and other classification techniques. For example, Gene Oracle is a software package that implements a deep learning model to classify biological samples using gene expression features as input 15 . In the Gene Oracle algorithm, expression profiles of candidate gene sets are tested for significant non-random classification potential of sample types (i.e.…”