This software is a Java-based tool that is designed for creating realistic in silico benchmarks for gene network algorithms. It is intended to provide an easy-to-use and customizable interface that allows users to generate accurate and reliable data for their research.
It is a known fact that the performance of reverse engineering methods can be influenced by the type of network structure they are applied to. This means that a method may not perform well on scale-free networks but may excel at Erdös-Rény networks. Due to this sensitivity to network structure, it is essential to have benchmark networks that possess a realistic structure for the comparison of methods to be fair.
Rather than using random graph models that only partially capture the structural nature of biological networks, GeneNetWeaver utilizes network structures extracted from known biological interaction networks. This approach generates realistic network structures that possess the appropriate properties needed for benchmarking.
Once the realistic network structures are established, GeneNetWeaver answers the next question by generating data from these networks using fit dynamical models. In previous in silico benchmarks such as the one conducted by Mendes et al., simple phenomenological models were utilized for this purpose. However, GeneNetWeaver has overcome this issue with the application of adequate dynamical models, enabling accurate and precise data generation.
Version 1.2: N/A