Automated analysis of biological oscillator models using mode decomposition

摘要: Motivation: Oscillating signals produced by biological systems have shapes, described their Fourier spectra, that can potentially reveal the mechanisms generate them. Extracting this information from measured is interesting for validation of theoretical models, discovery and classification interaction types, optimal experiment design. Results: An automated workflow analysis oscillating signals. A software package developed to match signal shapes hundreds a priori viable model structures defined class first-order differential equations. The computes parameter values each exploiting mode decomposition formulating matching problem in terms simultaneous polynomial On basis computed values, returns list models consistent with data. In tests synthetic datasets, it not only shortlists those used data but also shows excellent fits sometimes be achieved alternative listing all equations indicative how further invalidation might additional information. When applied microarray on mice, procedure finds several candidate describe interactions related circadian rhythm. This experimental oscillators indeed rich about gene regulation mechanisms. Availability: available at http://babylone.ulb.ac.be/autoosc/. Contact: tkonopka@ulb.ac.be Supplementary information:Supplementary are Bioinformatics online.