Abstract

Motivation: Understanding mechanisms and functions of microRNAs (miRNAs) is fundamental for the elucidation of many biological processes and of etiopathology of some diseases, such as tumors and neurodegenerative syndromes. We have developed a new biclustering algorithm, i.e. HOCCLUS2 [1], which is able to significantly correlate multiple miRNAs and their target genes to identify potential miRNA:mRNA regulatory networks. More recently, we developed a new probabilistic classifier [2] working in the semi-supervised ensemble learning setting, which allowed us to apply HOCCLUS2 on large-scale prediction data. In order to allow the researchers to exploit the obtained results, we have started to develop a web-based system, called mirNET, for the efficient query, retrieval, export, visualization and analysis of the discovered regulatory networks.Method: In [2], we presented a method which learns to combine the score of several prediction algorithms, in order to improve the reliability of the predicted interactions. The approach works in the semi-supervised ensemble learning setting which exploits information conveyed by both labeled (validated interactions, from miRTarBase) and unlabeled (predicted interactions, from mirDIP) instances. The algorithm HOCCLUS2 exploits the large set of produced predictions, with the associated probability, to extract a set of hierarchically organized biclusters. The construction of the hierarchy is performed by an iterative merging, considering both distance and density-based criteria.Extracted biclusters are also ranked on the basis of the p-values obtained by the Student's T-Test which compares intra- and inter- functional similarity of miRNA targets, computed on the basis of the gene classification provided in Gene Ontology (GO).mirNET database relies on PostgreSQL DBMS, while the web-based platform is built through the Play 2.2 Java framework and the Cytoscape library.Results: The mirNET database stores the set of interactions identified in [2] and the biclusters extracted by HOCCLUS2 from such set of interactions, with different parameters. In particular, mirNET stores approximately 5 million predicted interactions between 934 human miRNAs and 30,875 mRNAs, which are exploited in the construction of the hierarchies of biclusters representing potential miRNA regulatory networks.The mirNET web interface allows users to perform extraction and visualization of single interactions (with the score/probability assigned by the learning algorithm) and of biclusters of interest, as well as to easily browse whole biclusters hierarchies. Biclusters hierarchy browsing (i.e., navigation among parents and children biclusters) helps to identify intrinsic hierarchical organization of miRNAs in each specific context. The interface for the analysis of biclusters also provides a graph-based visualization of the predicted miRNA-gene interaction network. The database query system provides a series of filters to facilitate and re

Autore Pugliese

• D'elia Domenica

Tutti gli autori

• G. Pio ; M. Ceci ; D. D'Elia ; D. Malerba

Titolo volume/Rivista

Non Disponibile

Anno di pubblicazione

2014

ISSN

Non Disponibile

ISBN

Non Disponibile

Numero di citazioni Wos

Nessuna citazione

Ultimo Aggiornamento Citazioni

Non Disponibile

Numero di citazioni Scopus

Non Disponibile

Ultimo Aggiornamento Citazioni

Non Disponibile

Settori ERC

Non Disponibile

Codici ASJC

Non Disponibile