Microarray Gene Expression Analysis Using QR Decomposition, Machine Learning Algorithm, and Statistical Tool

The Allen Brain Atlas offers maps of gene expression and neuronal connectivity across the brain to help in the study of the molecular and functional architecture of the nervous system. Researchers have identified genes, cells, and structures with the help of this platform to employ several techniques to facilitate the analysis of microarray expression patterns and gene visualization. The first phase utilizes principal component analysis (PCA), the QR decomposition algorithm, and a nonlinear dimensionality reduction technique called t-SNE to reduce the number of dimensions. After t-SNE, the silhouette score and ward linkage were used to identify gene labels and optimal cluster values. In the next phase, we utilized a supervised learning approach, which involved using data and labels to assess the quality of the gene (feature). Furthermore, to measure the effectiveness of these features, we used Logistic Regression, which gave us an accuracy of 93.77%. We also used Tukey's HSD test and ANOVA for a more detailed assessment, revealing noteworthy gene expression differences among clusters. This comprehensive approach provides deeper insight into the genetic and connectivity patterns of the brain, which contributes to the broader field of neuroscientific research.