partCNV

partCNV infers locally aneuploid cells using single cell RNA-seq data through a statistical framework for rapid and accurate detection of aneuploid cells with local copy number deletion or amplification.

Key Features:

• EM algorithm with Poisson mixtures: Uses an EM algorithm with mixtures of Poisson distributions for classification of cells with copy number alterations.
• Cytogenetics integration: Incorporates cytogenetics information such as regional deletion or amplification to guide cell classification.
• Hidden Markov Model extension: Integrates a Hidden Markov Model for feature selection in the partCNVH variant to further improve accuracy when applicable.

Scientific Applications:

• Aneuploidy detection: Detects aneuploid cells with local copy number deletion or amplification from single cell RNA-seq data.
• Copy number variation analysis: Identifies regional copy number alterations including deletions and amplifications at the single cell level.

Methodology:

The method uses an EM algorithm with mixtures of Poisson distributions while incorporating cytogenetics information to guide classification. A Hidden Markov Model is integrated for feature selection when applicable.