Design of a Boosting-Based Similarity Measure for Evaluating Gene Expression Using Learning Approaches

A small number of the cell’s genes have expression levels that directly affect the functional or regulatory functions of the cell. Gene expression time series (GETS) keep track of each gene activity, which reveals underlying cellular dynamics. High-throughput GETS (HTGETS) investigations need the grouping of genes according to their temporal expression patterns, frequently done using unsupervised machine learning approaches. Nevertheless, most clustering methods either need to improve their ability to consider the temporal structure of the data or need to be improved by the short duration of time series for gene expression (GE). The innovative machine learning (ML)-based architecture known as boosting with similarity matrix (B-SM) provided can address these problems for grouping GETS and related difficulties. B-SM originally visualizes time series data to provide more detailed data interpretations. The produced pictures are then subjected to deep neural clustering. Studies using biological and the advantages of this innovative approach over traditional clustering techniques are shown via synthetic datasets. Additionally, we use an enrichment approach to show that the B-SM clusters are biologically plausible.