marginAlign

marginAlign aligns Oxford Nanopore Technologies' MinION long reads to reference genomes and calls single-nucleotide variants to enable accurate variant detection and resolution of complex genomic regions.

Key Features:

• Alignment Optimization: Optimized for aligning MinION two-dimensional (2D) reads to a reference genome, achieving mapping of over 99% of high-quality 2D reads with a reported mean identity of 85%.
• Error Rate Estimation: Uses expectation maximization to obtain maximum-likelihood estimates of insertion, deletion, and substitution error rates (reported as 4.9% insertions, 7.8% deletions, and 5.1% substitutions).
• Single-Nucleotide Variant Detection: Calls SNVs with up to 99% precision and recall by combining maximum-likelihood parameter estimates with marginalization over numerous possible read alignments.
• Resolution of Complex Genomic Regions: Leverages high-confidence alignments of long MinION reads to resolve complex regions, demonstrated by resolving copy number variation of the cancer-testis gene family CT47 in human chromosome Xq24.

Scientific Applications:

• Genomic Variant Analysis: Accurate SNV detection supports studies of genetic variation and its implications in disease.
• Structural Genomics: Long-read alignment and resolution of complex loci enable structural genomic analyses.
• Cancer Research: Resolution of copy number variation, including CT47 on Xq24, aids investigation of oncogenes and tumor suppressor gene alterations.

Methodology:

Aligns MinION 2D reads to a reference, applies expectation maximization to estimate maximum-likelihood insertion/deletion/substitution error rates, and performs SNV calling by marginalizing over possible read alignments using those parameter estimates.