Harmonic pulse analysis to detect biologically plausible gene regulatory networks

This paper presents Harmonic Pulse Analysis (HPA), an evaluation criteria and method to measure the “biological likeliness” of artificially generated gene regulatory networks, without the use of biological knowledge associated to genes, proteins and small molecules that constitute the target gene regulatory network. HPA is essential for the prediction of gene regulatory networks, which are mostly unknown. To evaluate the HPA, 1,000 networks were generated by successive modification of the gene regulatory network of Sea urchin endomesoderm, maintaining the number of genes and injection points. Additions, deletions and changes of links among genes were introduced to measure the variation of the networks' pulse density value, in order to extrapolate the discrimination threshold between truly biological network and approximatedly biological network. Generated networks were analyzed by the harmonic pulse analysis. 2% of systematically generated networks were within 1% margin of the harmonic pulse density of the Sea urchin early embryo network. 7% of networks were within 5% margin, and 28% within 10% margin.