User Guide

• 1. Independent Component Analysis
  • 1.1. Definition of ICA
    • 1.1.1. A simple example with the cocktail party problem
  • 1.2. FastICA solver
    • 1.2.1. Approximation of the negentropy
    • 1.2.2. Practical choice for the measuring function G
    • 1.2.3. A fixed-point algorithm
  • 1.3. Preconditioned ICA for Real Data (Picard) solver and the Infomax problem
    • 1.3.1. Maximum-likelihood approach and picard python package
• 2. Icasso algorithm
  • 2.1. Stability of the final components
  • 2.2. Choice of the number of components
• 3. Mutual Nearest Neighbors method
  • 3.1. The Reciprocal Best Hits method
  • 3.2. An illustration of the RBH method with gene expression data
  • 3.3. A natural extension of RBH: the mutual nearest neighbors method
• 4. Omics data analysis
  • 4.1. Omics data as linear mixture of latent biological signals
    • 4.1.1. A parallel with deconvolution of sound signals
  • 4.2. Enrichment analysis for annotating components
    • 4.2.1. Extraction of the top expressed genes
    • 4.2.2. Enrichment analysis with Reactome and ToppFun
  • 4.3. Single-cell data