Validation Pipeline for Computational Prediction of Genomics Annotations

Conference paper

DOI: 10.1007/978-3-319-44332-4_18

Part of the Lecture Notes in Computer Science book series (LNCS, volume 9874)
Cite this paper as:
Chicco D., Masseroli M. (2016) Validation Pipeline for Computational Prediction of Genomics Annotations. In: Angelini C., Rancoita P., Rovetta S. (eds) Computational Intelligence Methods for Bioinformatics and Biostatistics. CIBB 2015. Lecture Notes in Computer Science, vol 9874. Springer, Cham


Controlled biomolecular annotations are key concepts in computational genomics and proteomics, since they can describe the functional features of genes and their products in both a simple and computational way. Despite the importance of these annotations, many of them are missing, and the available ones contain errors and inconsistencies; furthermore, the discovery and validation of new annotations are very time-consuming tasks. For these reasons, recently many computer scientists developed several machine-learning algorithms able to computationally predict new gene-function relationships. While several of these methods have been easily adapted from different domains to bioinformatics, their validation remains a challenging aspect of a computational pipeline. Here, we propose a validation procedure based upon three different sub-phases, which is able to assess the precision of any algorithm predictions with a reliable degree of accuracy. We show some validation results obtained for Gene Ontology annotations of Homo sapiens genes that demonstrate the effectiveness of our validation approach.


Validation Gene Ontology Biomolecular annotations Receiver Operating Characteristic ROC curves Genomic and Proteomic Data Warehouse (GPDW) 

Copyright information

© Springer International Publishing Switzerland 2016

Authors and Affiliations

  1. 1.Princess Margaret Cancer CentreUniversity of TorontoOntarioCanada
  2. 2.Dipartimento di Elettronica Informazione e BioingegneriaPolitecnico di MilanoMilanItaly

Personalised recommendations