Dataset Information

Prediction and validation of gene-disease associations using methods inspired by social network analyses.

ABSTRACT: Correctly identifying associations of genes with diseases has long been a goal in biology. With the emergence of large-scale gene-phenotype association datasets in biology, we can leverage statistical and machine learning methods to help us achieve this goal. In this paper, we present two methods for predicting gene-disease associations based on functional gene associations and gene-phenotype associations in model organisms. The first method, the Katz measure, is motivated from its success in social network link prediction, and is very closely related to some of the recent methods proposed for gene-disease association inference. The second method, called Catapult (Combining dATa Across species using Positive-Unlabeled Learning Techniques), is a supervised machine learning method that uses a biased support vector machine where the features are derived from walks in a heterogeneous gene-trait network. We study the performance of the proposed methods and related state-of-the-art methods using two different evaluation strategies, on two distinct data sets, namely OMIM phenotypes and drug-target interactions. Finally, by measuring the performance of the methods using two different evaluation strategies, we show that even though both methods perform very well, the Katz measure is better at identifying associations between traits and poorly studied genes, whereas Catapult is better suited to correctly identifying gene-trait associations overall [corrected].

SUBMITTER: Singh-Blom UM

PROVIDER: S-EPMC3641094 | biostudies-literature | 2013

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Prediction and validation of gene-disease associations using methods inspired by social network analyses.

Singh-Blom U Martin UM Natarajan Nagarajan N Tewari Ambuj A Woods John O JO Dhillon Inderjit S IS Marcotte Edward M EM

PloS one 20130501 5

Correctly identifying associations of genes with diseases has long been a goal in biology. With the emergence of large-scale gene-phenotype association datasets in biology, we can leverage statistical and machine learning methods to help us achieve this goal. In this paper, we present two methods for predicting gene-disease associations based on functional gene associations and gene-phenotype associations in model organisms. The first method, the Katz measure, is motivated from its success in so ...[more]

PMID: 23650495

Similar Datasets

Project description:BackgroundHealth and social care systems were designed to be used primarily by people with single and acute diseases. However, a growing number of older adults are diagnosed with multiple long-term conditions (LTCs). The process of navigating the intricacies of health and social care systems to receive appropriate care presents significant challenges for older people living with multiple LTCs, which in turn can negatively influence their well-being and quality of life.ObjectiveThe long-term goal of this work is to design technology to assist people with LTCs in navigating health and social care systems. To do so, we must first understand how older people living with LTCs currently engage with and navigate their care networks. No published research describes and analyses the structure of formal and informal care networks of older adults with multiple LTCs, the frequency of interactions with each type of care service, and the problems that typically arise in these interactions.MethodsWe conducted a mixed-methods study and recruited 62 participants aged ?55 years who were living in England, had ?2 LTCs, and had completed a social network analysis questionnaire. Semistructured interviews were conducted with roughly a 10% subsample of the questionnaire sample: 4 women and 3 men. On average, interviewees aged 70 years and had 4 LTCs.ResultsPersonal care networks were complex and adapted to each individual. The task of building and subsequently navigating one's personal care network rested mainly on patients' shoulders. It was frequently the patients' task to bridge and connect the different parts of the system. The major factor leading to a satisfying navigation experience was found to be patients' assertive, determined, and proactive approaches. Furthermore, smooth communication and interaction between different parts of the care system led to more satisfying navigation experiences.ConclusionsTechnology to support care navigation for older adults with multiple LTCs needs to support patients in managing complex health and social care systems by effectively integrating the management of multiple conditions and facilitating communication among multiple stakeholders, while also offering flexibility to adapt to individual situations. Quality of care seems to be dependent on the determination and ability of patients. Those with less determination and fewer organization skills experience worse care. Thus, technology must aim to fulfill these coordination functions to ensure care is equitable across those who need it.

Project description:Modelling complex social behavior in the laboratory is challenging and requires analyses of dyadic interactions occurring over time in a physically and socially complex environment. In the current study, we approached the analyses of complex social interactions in group-housed male CD1 mice living in a large vivarium. Intensive observations of social interactions during a 3-week period indicated that male mice form a highly linear and steep dominance hierarchy that is maintained by fighting and chasing behaviors. Individual animals were classified as dominant, sub-dominant or subordinate according to their David's Scores and I& SI ranking. Using a novel dynamic temporal Glicko rating method, we ascertained that the dominance hierarchy was stable across time. Using social network analyses, we characterized the behavior of individuals within 66 unique relationships in the social group. We identified two individual network metrics, Kleinberg's Hub Centrality and Bonacich's Power Centrality, as accurate predictors of individual dominance and power. Comparing across behaviors, we establish that agonistic, grooming and sniffing social networks possess their own distinctive characteristics in terms of density, average path length, reciprocity out-degree centralization and out-closeness centralization. Though grooming ties between individuals were largely independent of other social networks, sniffing relationships were highly predictive of the directionality of agonistic relationships. Individual variation in dominance status was associated with brain gene expression, with more dominant individuals having higher levels of corticotropin releasing factor mRNA in the medial and central nuclei of the amygdala and the medial preoptic area of the hypothalamus, as well as higher levels of hippocampal glucocorticoid receptor and brain-derived neurotrophic factor mRNA. This study demonstrates the potential and significance of combining complex social housing and intensive behavioral characterization of group-living animals with the utilization of novel statistical methods to further our understanding of the neurobiological basis of social behavior at the individual, relationship and group levels.

Project description:BackgroundLittle is known about the relationship between miRNA and mRNA expression in Alzheimer's disease (AD) at early- or late-symptomatic stages. Sequence-based target prediction algorithms and anti-correlation profiles have been applied to predict miRNA targets using omics data, but this approach often leads to false positive predictions. Here, we applied the joint profiling analysis of mRNA and miRNA expression levels to Tg6799 AD model mice at 4 and 8 months of age using a network topology-based method. We constructed gene regulatory networks and used the PageRank algorithm to predict significant interactions between miRNA and mRNA.ResultsIn total, 8 cluster modules were predicted by the transcriptome data for co-expression networks of AD pathology. In total, 54 miRNAs were identified as being differentially expressed in AD. Among these, 50 significant miRNA-mRNA interactions were predicted by integrating sequence target prediction, expression analysis, and the PageRank algorithm. We identified a set of miRNA-mRNA interactions that were changed in the hippocampus of Tg6799 AD model mice. We determined the expression levels of several candidate genes and miRNA. For functional validation in primary cultured neurons from Tg6799 mice (MT) and littermate (LM) controls, the overexpression of ARRDC3 enhanced PPP1R3C expression. ARRDC3 overexpression showed the tendency to decrease the expression of miR139-5p and miR3470a in both LM and MT primary cells. Pathological environment created by Aβ treatment increased the gene expression of PPP1R3C and Sfpq but did not significantly alter the expression of miR139-5p or miR3470a. Aβ treatment increased the promoter activity of ARRDC3 gene in LM primary cells but not in MT primary cells.ConclusionsOur results demonstrate AD-specific changes in the miRNA regulatory system as well as the relationship between the expression levels of miRNAs and their targets in the hippocampus of Tg6799 mice. These data help further our understanding of the function and mechanism of various miRNAs and their target genes in the molecular pathology of AD.

Dataset Information

Prediction and validation of gene-disease associations using methods inspired by social network analyses.

Publications

Prediction and validation of gene-disease associations using methods inspired by social network analyses.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets