Validación de modelos genéticos en bioinformática: implementación y visualización

Abstract

Since the human genome was completely sequenced for the first time, the great scientific and technological advances in the biotechnology industry have greatly reduced the cost of experiments while significantly improving results. This has led to an exponential growth in the biological information available and, due to this huge amount of information, researchers are faced with mountains of data with only flakes of knowledge.

Approaches as Knowledge Database Discovery (KDD) are used to generate models that allows researcher to gain knowledge about complex biological systems.

Gene networks arose as a straightforward way of representing gene sets including their interactions. They are presented as a network structure where each node represents a gene or gene product (protein) while each edge denotes the relationship between the nodes at its ends. The concrete nature of each relationship and the meaning of its weight depend on the network architecture and the inference algorithm used. A gene network is an abstraction that facilitates the study of its underlying biological system. They are easy to visualize, and they are informative on their own.

Gene networks have been successfully used in clinical diagnosis and a large number of inferred interactions have been confirmed experimentally, thus confirming their reliability. The inference of gene networks has also allowed a better understanding of fundamental processes that occur in living organisms such as development or nutrition and metabolic coordination.

Research has focused on inferring these networks using different experimental and computational techniques, as well as analyzing those networks to extract knowledge. Also, a significant number of methods have been developed to validate the inferred networks in order to verify their quality and reliability.

All the methodologies of gene network inference, analysis, and validation are based on algorithms and computer tools. Given the increasing importance and popularity of these computational approaches, it becomes increasingly critical to ensure that the software is usable and accessible, as these features provide the basis for the reproducibility of published biomedical research.

Based on the existing need for automatic techniques of inference, analysis and validation of models for the study of interactions between genes and the deficiencies in existing techniques, this work presents different novel approaches for the inference, analysis and validation of genetic models, especially gene networks, with a special emphasis on the usability and accessibility of the proposed solutions.