Single Cell Genotyper (SCG)

Single Cell Genotyper (SCG) infers clonal genotypes and population structure from single-cell DNA sequencing data to characterize tumor clonal architecture.

Key Features:

• Statistical model: Implements a probabilistic statistical model to represent genotypes and clonal mixtures.
• Mean-field variational inference: Uses a mean-field variational inference method for parameter and latent-state estimation.
• Inference of clonal genotypes: Produces inferred clonal genotypes from single-cell sequencing observations.
• Clustering of single-cell (nucleus) data: Performs clustering of single-cell (nucleus) data to identify distinct clones.
• Handling missing values: Accounts for missing values commonly observed in single-cell genomic data.
• Correction for biased allelic counts: Models and mitigates biased allelic counts in single-cell DNA sequencing.
• Detection of false genotype measurements: Models false genotype measurements that can arise from sequencing of multiple cells.
• Probabilistic models and machine learning: Integrates probabilistic models and machine learning algorithms for robust genotype and cluster inference.

Scientific Applications:

• Cancer clonal architecture analysis: Characterizes clonal genotypes and the population structure within human cancers.
• Study of tumor evolution: Enables exploration of evolutionary dynamics and tumorigenesis at single-cell resolution.
• Mutation tracking over time: Facilitates tracking of genetic mutations across cells and over time.
• Identification of therapeutic targets: Supports identification of potential therapeutic targets by resolving clonal genotypes.

Methodology:

SCG applies a statistical model integrated with mean-field variational inference and employs probabilistic models and machine learning algorithms to infer clonal genotypes and cluster single-cell (nucleus) data.