Multi-scale genetic network inference based on time series gene expression profiles

摘要: This work integrates multi-scale clustering and short-time correlation to estimate genetic networks with different time resolutions detail levels. Gene expression data are noisy large scale. Clustering is widely used group genes similar pattern. The cluster centers can be infer the among these clusters. introduces Multi-scale Fuzzy K-means algorithm uncover groups of coregulated capture in levels detail. Time series profiles provide dynamic information for inferring gene regulatory relationships. Large scale network inference, identifying transient interactions feedback loops as well differentiating direct indirect major challenges inference. Time delay edge direction. Partial directed-separation theory help differentiate identify loops. constraint-based time-correlation (CBTC) inference that combines methods estimation more fully characterize networks. regulation happen specific periods conditions instead across whole profile. Short-time interactions. The discovery was mainly validated using yeast cell cycle data. successfully identified development stages, negative loops, indicated how dynamically changes over time. inferred reflect most previously by genome-wide location analysis match extant literature. At detailed level, about (or clusters) them. Interesting genes, clusters were identified, which literature ontology hypotheses further studies.