HTRgene: Integrating Multiple Heterogeneous Time-series Data to Investigate Cold and Heat Stress Response Signaling Genes in Arabidopsis

Time-series gene expression datasets measured under the same stress are increasing, and they are available for an integrated analysis to investigate stress response signaling genes. However, the integrated analysis requires well-established and strategic methods because meta-properties (the number of time points and phenotype) are heterogeneous across multiple samples. We present an algorithm, HTRgene, that performs the integrated analysis of multiple heterogeneous time-series datasets under the same stress condition. HTRgene identifies differentially expressed genes (DEGs) that are consistently differentially expressed across multiple samples, clusters the genes based on co-expression patterns, detects the response times of the gene clusters, and then determines the consistent response order of the clusters across multiple samples to produce response order preserving DEGs. In experiments using 28 and 24 timeseries sample gene expression datasets under cold and heat stress, HTRgene successfully reproduced biological mechanisms of cold and heat stress in Arabidopsis, which are well documented in the literature. We also show that HTRgene can identify candidate response genes with higher accuracy than existing tools. We believe that HTRgene will be very useful in investigating biological mechanisms of stress response through multiple time-series data integration analysis. HTRgene is available at https://bhi-kimlab.github.io/projects/#HTRgene.