Reconstruction of gene regulatory networks with multi-objective particle swarm optimisers
- PDF / 2,387,940 Bytes
- 20 Pages / 595.224 x 790.955 pts Page_size
- 81 Downloads / 158 Views
Reconstruction of gene regulatory networks with multi-objective particle swarm optimisers Sandro Hurtado1 · Jose´ Garc´ıa-Nieto1
· Ismael Navas-Delgado1 · Antonio J. Nebro1 · Jose´ F. Aldana-Montes1
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract The computational reconstruction of Gene Regulatory Networks (GRNs) from gene expression data has been modelled as a complex optimisation problem, which enables the use of sophisticated search methods to address it. Among these techniques, particle swarm optimisation based algorithms stand out as prominent techniques with fast convergence and accurate network inferences. A multi-objective approach for the inference of GRNs consists of optimising a given network’s topology while tuning the kinetic order parameters in an S-System, thus preventing the use of unnecessary penalty weights and enables the adoption of Pareto optimality based algorithms. In this study, we empirically assess the behaviour of a set of multi-objective particle swarm optimisers based on different archiving and leader selection strategies in the scope of the inference of GRNs. The main goal is to provide system biologists with experimental evidence about which optimisation technique performs with higher success for the inference of consistent GRNs. The experiments conducted involve time-series datasets of gene expression taken from the DREAM3/4 standard benchmarks, as well as in vivo datasets from IRMA and Melanoma cancer samples. Our study shows that multi-objective particle swarm optimiser OMOPSO obtains the best overall performance. Inferred networks show biological consistency in accordance with in vivo studies in the literature. Keywords Multi-objective optimisation · Particle swarm optimisation · Gene regulatory networks · Performance and quality analysis · Biological systems
1 Introduction In the last years, multiple optimisation techniques such as evolutionary algorithms [16, 17, 30], and especially particle swarm optimisation [1, 21, 24, 53, 60], have been applied to the inference of Gene Regulatory Networks (GRNs) from gene expression time-series. This is a complex problem found in computational biology [19], which This work has been partially funded by the Spanish Ministry of Science and Innovation via Grant TIN2017-86049-R (AEI/FEDER, UE) and Andalusian PAIDI program with grant P18-RT-2799. Jos´e Garc´ıa-Nieto is the recipient of a Post-Doctoral fellowship of “Captaci´on de Talento para la Investigaci´on” Plan Propio at Universidad de M´alaga. Jos´e Garc´ıa-Nieto
[email protected] 1
Institute for Software Technologies and Software Engineering (ITIS), Biomedical Research Institute of M´alaga (IBIMA), Department of Computer Languages and Computing Sciences, University of M´alaga ETSI Inform´atica, Campus de Teatinos, M´alaga, 29071, Spain
consists of tuning parameters of a model that quantitatively reproduces the dynamics of a given biological system. In this regard, there exists advanced computational models capable of inferring the topology of gene inte
Data Loading...
