SoloDel

SoloDel identifies somatic deletions from whole-genome sequencing data, distinguishing low-frequency somatic events from germline variations and enabling analysis without matched control samples.

Key Features:

• Probabilistic Somatic Mutation Progression Model: Incorporates a probabilistic model describing the occurrence and propagation of somatic mutations within cellular lineages to aid distinction between somatic and germline deletions.
• Gaussian Mixture Modeling: Models mixed populations of somatic and germline deletions using a Gaussian mixture model applied to read-depth ratios at loci with discordant reads, with parameters estimated via the expectation-maximization algorithm.
• Integration with Structural Variation Callers: Integrates outputs from conventional structural variation callers to improve classification of somatic versus germline deletions.
• Performance Without Matched Controls: Maintains high performance across mutated subpopulation sizes of 10–70% even in the absence of matched normal samples.
• Validation and Application: Validated against experimentally confirmed somatic deletions from neuropsychiatric whole-genome sequencing data.

Scientific Applications:

• Cancer genomics: Identification and characterization of somatic deletions in cancer whole-genome sequencing studies.
• Neurogenetics: Detection of somatic deletions in neuropsychiatric whole-genome sequencing datasets.
• Studies of heterogeneous or limited samples: Analysis of somatic deletions in heterogeneous cell populations or studies lacking matched normal samples.

Methodology:

Uses a probabilistic somatic mutation progression model and Gaussian mixture models on read-depth ratios at loci with discordant reads, with parameter estimation via the expectation-maximization algorithm and integration of conventional structural variation caller outputs.