SMAP

SMAP applies discrete-index Hidden Markov Models (HMMs) to array-based comparative genomic hybridization (array-CGH) data to assign DNA copy numbers to chromosomal regions for accurate genome-wide DNA copy number profiling.

Key Features:

• Hidden Markov Modeling: SMAP employs discrete-index HMMs to model DNA copy number variations and probabilistically assign copy-number states.
• Incorporation of Genomic Overlap and Distance: SMAP integrates genomic overlap between clones and genomic distance into the model to account for heterogeneity in clone overlap.
• User-Controllable Parameterization: The method accepts a priori knowledge via adjustable parameters (priors) to tailor sensitivity to deviations of varying lengths and amplitudes.
• Genome-Wide Model Inference: SMAP infers model parameters at the genome-wide scale rather than on a chromosome-by-chromosome basis to reduce the risk of overfitting.
• Superior Performance on Synthetic Data: Comparative analyses on synthetic data sets demonstrate that SMAP outperforms recent methods and recognizes both large-scale regions with abnormal copy number and changes affecting single features.

Scientific Applications:

• Array-CGH copy number profiling: Generation of genome-wide DNA copy number profiles from array-CGH data.
• Detection of genetic aberrations in disease: Identification and characterization of genetic aberrations associated with diseases such as cancer.
• Resolution of focal and broad events: Detection of both large-scale copy number alterations and single-feature (focal) copy number changes.

Methodology:

SMAP uses discrete-index HMMs to assign DNA copy numbers to chromosomal regions, integrates genomic overlap and genomic distance into the HMM, performs genome-wide parameter inference, allows user-specified prior parameterization, and was evaluated by comparative analyses on synthetic data sets.