Spatiotemporal Relationships Extraction from ISH Gene Expression Data
Résumé
Imperfection is a common feature in almost all real-world data, although it usually hides crucial knowledge of major interest. Using theories of uncertainty to model gene expression (gen-exp) areas in the embryo can help extract hidden relationships between genes while taking into account possible imprecision in the boundaries of gen-exp zones and their nuanced strength. This study proposes fuzzy and possibilistic modeling of spatiotemporal data from In-Situ Hybridization (ISH) sequences of images representing gen-exp areas in different embryonic developmental phases of the model species Edinburgh Mouse. Following a series of preprocessing steps on these images to improve feature extraction, an adaptation of the Apriori algorithm to the fuzzy and possibilistic logic is proposed for mining two types of Association Rules (AR) that represent the spatial correlations between gen-exp areas in the embryo, and the temporal correlations as well as the relationships between genes that co-express in these ISH image sequences. Finally, an analysis of extracted spatiotemporal fuzzy and possibilistic AR is provided to decide which modeling is the most suitable. Biological interpretation of the results obtained confirms their adequacy with the domain principles. The extracted knowledge can help biologists better understand the interactions between genes, discover the coexpressed sets of genes that have the same functional role, and model normal gen-exp for detecting abnormal gen-exp that can cause genetic diseases.
Citer ce document
Accès au document
Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter
Voir l'article sur le site de la revueAuteur(s)
Statistiques
Consultations : 1
Téléchargements : 0