Optimizationof neural network architecture using genetic programming improvesdetection and modeling of gene-gene interactions in studies of humandiseases

Background

Appropriate definitionof neural network architecture prior to data analysis is crucialfor successful data mining. This can be challenging when the underlyingmodel of the data is unknown. The goal of this study was to determinewhether optimizing neural network architecture using genetic programmingas a machine learning strategy would improve the ability of neural networksto model and detect nonlinear interactions among genes in studiesof common human diseases.

Results

Using simulateddata, we show that a genetic programming optimized neural network approachis able to model gene-gene interactions as well as a traditionalback propagation neural network. Furthermore, the genetic programmingoptimized neural network is better than the traditional back propagationneural network approach in terms of predictive ability and powerto detect gene-gene interactions when non-functional polymorphismsare present.

Conclusion

This study suggeststhat a machine learning strategy for optimizing neural network architecturemay be preferable to traditional trial-and-error approaches forthe identification and characterization of gene-gene interactionsin common, complex human diseases.

     

A General Model for Multilocus Epistatic Interactions in Case-Control Studies

Background

Epistasis, i.e., the interaction of alleles at different loci, is thought to play a central role in the formation and progression of complex diseases. The complexity of disease expression should arise from a complex network of epistatic interactions involving multiple genes.

Methodology

We develop a general model for testing high-order epistatic interactions for a complex disease in a case-control study. We incorporate the quantitative genetic theory of high-order epistasis into the setting of cases and controls sampled from a natural population. The new model allows the identification and testing of epistasis and its various genetic components.

Conclusions

Simulation studies were used to examine the power and false positive rates of the model under different sampling strategies. The model was used to detect epistasis in a case-control study of inflammatory bowel disease, in which five SNPs at a candidate gene were typed, leading to the identification of a significant three-locus epistasis.     

New application of intelligent agents in sporadic amyotrophic lateral sclerosis identifies unexpected specific genetic background

Background  

Few genetic factors predisposing to the sporadic form of amyotrophic lateral sclerosis (ALS) have been identified, but the pathology itself seems to be a true multifactorial disease in which complex interactions between environmental and genetic susceptibility factors take place. The purpose of this study was to approach genetic data with an innovative statistical method such as artificial neural networks to identify a possible genetic background predisposing to the disease. A DNA multiarray panel was applied to genotype more than 60 polymorphisms within 35 genes selected from pathways of lipid and homocysteine metabolism, regulation of blood pressure, coagulation, inflammation, cellular adhesion and matrix integrity, in 54 sporadic ALS patients and 208 controls. Advanced intelligent systems based on novel coupling of artificial neural networks and evolutionary algorithms have been applied. The results obtained have been compared with those derived from the use of standard neural networks and classical statistical analysis     

Distilling a Visual Network of Retinitis Pigmentosa Gene-Protein Interactions to Uncover New Disease Candidates

Background

Retinitis pigmentosa (RP) is a highly heterogeneous genetic visual disorder with more than 70 known causative genes, some of them shared with other non-syndromic retinal dystrophies (e.g. Leber congenital amaurosis, LCA). The identification of RP genes has increased steadily during the last decade, and the 30% of the cases that still remain unassigned will soon decrease after the advent of exome/genome sequencing. A considerable amount of genetic and functional data on single RD genes and mutations has been gathered, but a comprehensive view of the RP genes and their interacting partners is still very fragmentary. This is the main gap that needs to be filled in order to understand how mutations relate to progressive blinding disorders and devise effective therapies.

Methodology

We have built an RP-specific network (RPGeNet) by merging data from different sources: high-throughput data from BioGRID and STRING databases, manually curated data for interactions retrieved from iHOP, as well as interactions filtered out by syntactical parsing from up-to-date abstracts and full-text papers related to the RP research field. The paths emerging when known RP genes were used as baits over the whole interactome have been analysed, and the minimal number of connections among the RP genes and their close neighbors were distilled in order to simplify the search space.

Conclusions

In contrast to the analysis of single isolated genes, finding the networks linking disease genes renders powerful etiopathological insights. We here provide an interactive interface, RPGeNet, for the molecular biologist to explore the network centered on the non-syndromic and syndromic RP and LCA causative genes. By integrating tissue-specific expression levels and phenotypic data on top of that network, a more comprehensive biological view will highlight key molecular players of retinal degeneration and unveil new RP disease candidates.     

Artificial Neural Network Inference (ANNI): A Study on Gene-Gene Interaction for Biomarkers in Childhood Sarcomas

Objective

To model the potential interaction between previously identified biomarkers in children sarcomas using artificial neural network inference (ANNI).

Method

To concisely demonstrate the biological interactions between correlated genes in an interaction network map, only 2 types of sarcomas in the children small round blue cell tumors (SRBCTs) dataset are discussed in this paper. A backpropagation neural network was used to model the potential interaction between genes. The prediction weights and signal directions were used to model the strengths of the interaction signals and the direction of the interaction link between genes. The ANN model was validated using Monte Carlo cross-validation to minimize the risk of over-fitting and to optimize generalization ability of the model.

Results

Strong connection links on certain genes (TNNT1 and FNDC5 in rhabdomyosarcoma (RMS); FCGRT and OLFM1 in Ewing’s sarcoma (EWS)) suggested their potency as central hubs in the interconnection of genes with different functionalities. The results showed that the RMS patients in this dataset are likely to be congenital and at low risk of cardiomyopathy development. The EWS patients are likely to be complicated by EWS-FLI fusion and deficiency in various signaling pathways, including Wnt, Fas/Rho and intracellular oxygen.

Conclusions

The ANN network inference approach and the examination of identified genes in the published literature within the context of the disease highlights the substantial influence of certain genes in sarcomas.     

Bayesian neural networks for detecting epistasis in genetic association studies

Background

Discovering causal genetic variants from large genetic association studies poses many difficult challenges. Assessing which genetic markers are involved in determining trait status is a computationally demanding task, especially in the presence of gene-gene interactions.

Results

A non-parametric Bayesian approach in the form of a Bayesian neural network is proposed for use in analyzing genetic association studies. Demonstrations on synthetic and real data reveal they are able to efficiently and accurately determine which variants are involved in determining case-control status. By using graphics processing units (GPUs) the time needed to build these models is decreased by several orders of magnitude. In comparison with commonly used approaches for detecting interactions, Bayesian neural networks perform very well across a broad spectrum of possible genetic relationships.

Conclusions

The proposed framework is shown to be a powerful method for detecting causal SNPs while being computationally efficient enough to handle large datasets.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-014-0368-0) contains supplementary material, which is available to authorized users.     

Network properties of human disease genes with pleiotropic effects

Background  

The ability of a gene to cause a disease is known to be associated with the topological position of its protein product in the molecular interaction network. Pleiotropy, in human genetic diseases, refers to the ability of different mutations within the same gene to cause different pathological effects. Here, we hypothesized that the ability of human disease genes to cause pleiotropic effects would be associated with their network properties.     

Molecular mechanistic associations of human diseases

Background  

The study of relationships between human diseases provides new possibilities for biomedical research. Recent achievements on human genetic diseases have stimulated interest to derive methods to identify disease associations in order to gain further insight into the network of human diseases and to predict disease genes.     

A surrogate-based approach for post-genomic partner identification

Background  

Modern drug discovery is concerned with identification and validation of novel protein targets from among the 30,000 genes or more postulated to be present in the human genome. While protein-protein interactions may be central to many disease indications, it has been difficult to identify new chemical entities capable of regulating these interactions as either agonists or antagonists.     

A module-based analytical strategy to identify novel disease-associated genes shows an inhibitory role for interleukin 7 Receptor in allergic inflammation

Background  

The identification of novel genes by high-throughput studies of complex diseases is complicated by the large number of potential genes. However, since disease-associated genes tend to interact, one solution is to arrange them in modules based on co-expression data and known gene interactions. The hypothesis of this study was that such a module could be a) found and validated in allergic disease and b) used to find and validate one ore more novel disease-associated genes.     

Network Properties of Complex Human Disease Genes Identified through Genome-Wide Association Studies

Background

Previous studies of network properties of human disease genes have mainly focused on monogenic diseases or cancers and have suffered from discovery bias. Here we investigated the network properties of complex disease genes identified by genome-wide association studies (GWAs), thereby eliminating discovery bias.

Principal findings

We derived a network of complex diseases (n = 54) and complex disease genes (n = 349) to explore the shared genetic architecture of complex diseases. We evaluated the centrality measures of complex disease genes in comparison with essential and monogenic disease genes in the human interactome. The complex disease network showed that diseases belonging to the same disease class do not always share common disease genes. A possible explanation could be that the variants with higher minor allele frequency and larger effect size identified using GWAs constitute disjoint parts of the allelic spectra of similar complex diseases. The complex disease gene network showed high modularity with the size of the largest component being smaller than expected from a randomized null-model. This is consistent with limited sharing of genes between diseases. Complex disease genes are less central than the essential and monogenic disease genes in the human interactome. Genes associated with the same disease, compared to genes associated with different diseases, more often tend to share a protein-protein interaction and a Gene Ontology Biological Process.

Conclusions

This indicates that network neighbors of known disease genes form an important class of candidates for identifying novel genes for the same disease.     

Gene Prospector: An evidence gateway for evaluating potential susceptibility genes and interacting risk factors for human diseases

Background  

Millions of single nucleotide polymorphisms have been identified as a result of the human genome project and the rapid advance of high throughput genotyping technology. Genetic association studies, such as recent genome-wide association studies (GWAS), have provided a springboard for exploring the contribution of inherited genetic variation and gene/environment interactions in relation to disease. Given the capacity of such studies to produce a plethora of information that may then be described in a number of publications, selecting possible disease susceptibility genes and identifying related modifiable risk factors is a major challenge. A Web-based application for finding evidence of such relationships is key to the development of follow-up studies and evidence for translational research.     

A synthetic three-color scaffold for monitoring genetic regulation and noise

Background  

Current methods for analyzing the dynamics of natural regulatory networks, and quantifying synthetic circuit function, are limited by the lack of well-characterized genetic measurement tools. Fluorescent reporters have been used to measure dynamic gene expression, but recent attempts to monitor multiple genes simultaneously in single cells have not focused on independent, isolated measurements. Multiple reporters can be used to observe interactions between natural genes, or to facilitate the 'debugging' of biologically engineered genetic networks. Using three distinguishable reporter genes in a single cell can reveal information not obtainable from only one or two reporters. One application of multiple reporters is the use of genetic noise to reveal regulatory connections between genes. Experiments in both natural and synthetic systems would benefit from a well-characterized platform for expressing multiple reporter genes and synthetic network components.     

Prediction of drugs having opposite effects on disease genes in a directed network

Background

Developing novel uses of approved drugs, called drug repositioning, can reduce costs and times in traditional drug development. Network-based approaches have presented promising results in this field. However, even though various types of interactions such as activation or inhibition exist in drug-target interactions and molecular pathways, most of previous network-based studies disregarded this information.

Methods

We developed a novel computational method, Prediction of Drugs having Opposite effects on Disease genes (PDOD), for identifying drugs having opposite effects on altered states of disease genes. PDOD utilized drug-drug target interactions with ‘effect type’, an integrated directed molecular network with ‘effect type’ and ‘effect direction’, and disease genes with regulated states in disease patients. With this information, we proposed a scoring function to discover drugs likely to restore altered states of disease genes using the path from a drug to a disease through the drug-drug target interactions, shortest paths from drug targets to disease genes in molecular pathways, and disease gene-disease associations.

Results

We collected drug-drug target interactions, molecular pathways, and disease genes with their regulated states in the diseases. PDOD is applied to 898 drugs with known drug-drug target interactions and nine diseases. We compared performance of PDOD for predicting known therapeutic drug-disease associations with the previous methods. PDOD outperformed other previous approaches which do not exploit directional information in molecular network. In addition, we provide a simple web service that researchers can submit genes of interest with their altered states and will obtain drugs seeming to have opposite effects on altered states of input genes at http://gto.kaist.ac.kr/pdod/index.php/main.

Conclusions

Our results showed that ‘effect type’ and ‘effect direction’ information in the network based approaches can be utilized to identify drugs having opposite effects on diseases. Our study can offer a novel insight into the field of network-based drug repositioning.

     

Candidate gene prioritization by network analysis of differential expression using machine learning approaches

Background  

Discovering novel disease genes is still challenging for diseases for which no prior knowledge - such as known disease genes or disease-related pathways - is available. Performing genetic studies frequently results in large lists of candidate genes of which only few can be followed up for further investigation. We have recently developed a computational method for constitutional genetic disorders that identifies the most promising candidate genes by replacing prior knowledge by experimental data of differential gene expression between affected and healthy individuals.     

Differences in Gene-Gene Interactions in Graves’ Disease Patients Stratified by Age of Onset

Background

Graves’ disease (GD) is a complex disease in which genetic predisposition is modified by environmental factors. Each gene exerts limited effects on the development of autoimmune disease (OR = 1.2–1.5). An epidemiological study revealed that nearly 70% of the risk of developing inherited autoimmunological thyroid diseases (AITD) is the result of gene interactions. In the present study, we analyzed the effects of the interactions of multiple loci on the genetic predisposition to GD. The aim of our analyses was to identify pairs of genes that exhibit a multiplicative interaction effect.

Material and Methods

A total of 709 patients with GD were included in the study. The patients were stratified into more homogeneous groups depending on the age at time of GD onset: younger patients less than 30 years of age and older patients greater than 30 years of age. Association analyses were performed for genes that influence the development of GD: HLADRB1, PTPN22, CTLA4 and TSHR. The interactions among polymorphisms were analyzed using the multiple logistic regression and multifactor dimensionality reduction (MDR) methods.

Results

GD patients stratified by the age of onset differed in the allele frequencies of the HLADRB1*03 and 1858T polymorphisms of the PTPN22 gene (OR = 1.7, p = 0.003; OR = 1.49, p = 0.01, respectively). We evaluated the genetic interactions of four SNPs in a pairwise fashion with regard to disease risk. The coexistence of HLADRB1 with CTLA4 or HLADRB1 with PTPN22 exhibited interactions on more than additive levels (OR = 3.64, p = 0.002; OR = 4.20, p < 0.001, respectively). These results suggest that interactions between these pairs of genes contribute to the development of GD. MDR analysis confirmed these interactions.

Conclusion

In contrast to a single gene effect, we observed that interactions between the HLADRB1/PTPN22 and HLADRB1/CTLA4 genes more closely predicted the risk of GD onset in young patients.     

Midkine is a NF-κB-inducible gene that supports prostate cancer cell survival

Background

Atherosclerotic peripheral arterial disease (PAD) affects 8–10 million people in the United States and is associated with a marked impairment in quality of life and an increased risk of cardiovascular events. Noninvasive assessment of PAD is performed by measuring the ankle-brachial index (ABI). Complex traits, such as ABI, are influenced by a large array of genetic and environmental factors and their interactions. We attempted to characterize the genetic architecture of ABI by examining the main and interactive effects of individual single nucleotide polymorphisms (SNPs) and conventional risk factors.

Methods

We applied linear regression analysis to investigate the association of 435 SNPs in 112 positional and biological candidate genes with ABI and related physiological and biochemical traits in 1046 non-Hispanic white, hypertensive participants from the Genetic Epidemiology Network of Arteriopathy (GENOA) study. The main effects of each SNP, as well as SNP-covariate and SNP-SNP interactions, were assessed to investigate how they contribute to the inter-individual variation in ABI. Multivariable linear regression models were then used to assess the joint contributions of the top SNP associations and interactions to ABI after adjustment for covariates. We reduced the chance of false positives by 1) correcting for multiple testing using the false discovery rate, 2) internal replication, and 3) four-fold cross-validation.

Results

When the results from these three procedures were combined, only two SNP main effects in NOS3, three SNP-covariate interactions (ADRB2 Gly 16 – lipoprotein(a) and SLC4A5 – diabetes interactions), and 25 SNP-SNP interactions (involving SNPs from 29 different genes) were significant, replicated, and cross-validated. Combining the top SNPs, risk factors, and their interactions into a model explained nearly 18% of variation in ABI in the sample. SNPs in six genes (ADD2, ATP6V1B1, PRKAR2B, SLC17A2, SLC22A3, and TGFB3) were also influencing triglycerides, C-reactive protein, homocysteine, and lipoprotein(a) levels.

Conclusion

We found that candidate gene SNP main effects, SNP-covariate and SNP-SNP interactions contribute to the inter-individual variation in ABI, a marker of PAD. Our findings underscore the importance of conducting systematic investigations that consider context-dependent frameworks for developing a deeper understanding of the multidimensional genetic and environmental factors that contribute to complex diseases.     

Function approximation approach to the inference of reduced NGnet models of genetic networks

Background  

The inference of a genetic network is a problem in which mutual interactions among genes are deduced using time-series of gene expression patterns. While a number of models have been proposed to describe genetic regulatory networks, this study focuses on a set of differential equations since it has the ability to model dynamic behavior of gene expression. When we use a set of differential equations to describe genetic networks, the inference problem can be defined as a function approximation problem. On the basis of this problem definition, we propose in this study a new method to infer reduced NGnet models of genetic networks.