JACOBI4

JACOBI4 performs multivariate analysis of complex biological data by converting non-numeric descriptions such as genetic sequences, images, and similarity matrices into object similarity matrices and applying principal coordinates analysis, multidimensional scaling, and bootstrap methods to analyze high-dimensional datasets.

Key Features:

• Dimensionality Reduction: Transforms large-scale variables into object similarity matrices to reduce dimensionality to manageable levels (hundreds or thousands).
• Principal Coordinates Analysis (PCo) and Multidimensional Scaling (MDS): Implements PCo and MDS for analysis of dissimilarity matrices without relying on traditional variable-based representations.
• Bootstrap Methodology: Incorporates B. Efron’s bootstrap resampling to estimate statistical confidence via repeated analyses ranging from hundreds to millions of iterations, with support for parallel processing.
• Automation and Scripting: Provides scripting capabilities and pre-developed scripts to automate repeated multivariate and bootstrap analyses.

Scientific Applications:

• Gene expression analysis: Supports multivariate analysis of gene expression datasets, including studies related to different diseases.
• Molecular sequence variability: Enables analysis of molecular sequence variability using similarity-based multivariate methods on genetic sequences.
• Mass spectrometry data analysis: Applicable to mass spectrometry datasets characterized by high dimensionality and lack of explicit variable definitions.
• Genome research: Facilitates analysis in genome research and comparative genomics using sequence similarity matrices and high-dimensional data approaches.

Methodology:

Computational methods explicitly include conversion of data to object similarity/dissimilarity matrices, principal coordinates analysis (PCo), multidimensional scaling (MDS), bootstrap resampling (per B. Efron), repeated analyses across hundreds to millions of iterations, and parallel processing.