Efficient Biclustering Algorithms for Identifying Transcriptional Regulation Relationships Using Time Series Gene Expression Data

Biclustering algorithms have shown to be remarkably effective in a variety of applications. Although the biclustering problem is known to be NP-complete, in the particular case of time series gene expression data analysis, efficient and complete biclustering algorithms, are known and have been used to identify biologically relevant expression patterns. However, these algorithms, namely CCC-Biclustering (a linear time algorithm to identify biclusters with perfect expression patterns) and e-CCC-Biclustering (a polynomial time algorithm to identify biclusters with approximate expression patterns) do not take into account the importance of time-lagged and anti-correlation relationships in the study of transcription regulation using time series expression data. Furthermore, it is known that certain type of network motifs can generate temporal programs of expression, in which genes are activated one by one in a predefined order and can thus produce time-lagged expression patterns. In this context, we proposed extensions to these state of the art algorithms to analyze time series gene expression data in order to identify time-lagged activation together with anti-correlation while dealing directly with missing values. We present preliminary results obtained by applying the extended version of the CCC-Biclustering algorithm to the transcriptomic expression patterns occurring in Saccharomyces cerevisiae in response to heat stress.