NGSoptwin

NGSoptwin estimates the optimal genomic window size for analyzing very low-coverage next-generation sequencing (NGS) data (<0.1×) to improve detection of copy-number alterations and other genomic features.

Key Features:

• Optimal Window Size Estimation: Identifies window sizes that balance sparse zero-count bins and excessive smoothing to retain detectable genomic features.
• Model Assumptions: Assumes a step-function model for read density across the genome.
• Statistical Methodology: Evaluates window sizes using Akaike's Information Criterion (AIC) and cross-validation (CV) log-likelihood and plots these criteria against window size to select the minimum AIC or maximum CV log-likelihood.
• Low-Coverage NGS Handling: Targets very low-coverage NGS datasets (<0.1×) where many bins may contain zero counts.
• Implementation: Implemented as an R package for analysis of sequencing read-count data.

Scientific Applications:

• Copy-Number Alteration Analysis: Optimizes window size to improve detection and evaluation of copy-number alterations in genomic data.
• Tumour Sample Analysis: Applied to tumour sequencing datasets to extract genomic information from low-coverage cancer samples.
• Simulated Data Validation: Validated using simulated datasets to assess performance across different scenarios.

Methodology:

Uses a step-function read-density model and assesses candidate window sizes by computing and plotting Akaike's Information Criterion (AIC) and cross-validation (CV) log-likelihood across window sizes, selecting the window that minimizes AIC or maximizes CV log-likelihood.