pCMF

Statistics
R
Genomics
Probabilistic count matrix factorization for single cell transcriptomic data analyses (dimension reduction, visualization).
Author

Ghislain Durif

Published

January 15, 2015

Publication Code (package) Code (experiments)

(full development)

Collaboration with:

  • Sophie Lambert-Lacroix (Univ Grenoble Alpes)
  • Franck Picard (CNRS - LBBE - Univ Lyon)

The pCMF package contains mplementation of the probabilistic Count Matrix Factorization (pCMF) method based on the Gamma-Poisson hirerarchical factor model with potential sparisty-inducing priors on factor V. This method is specifically designed to analyze large count matrices with numerous potential drop-out events (also called zero-inflation) such as gene expression profiles from single cells data (scRNA-seq) obtained by high throughput sequencing.

Programming:

  • R
  • C++

Keywords

  • Statistics
  • Unsupervised learning
  • Data visualization
  • Dimension reduction
  • High-dimensional data
  • Probabilistic matrix factorization
  • Genomics
  • Single-cell
  • RNA-seq

Project:

  • ABS4NGS