MAQ

MAQ aligns short reads from next-generation sequencing (e.g., Illumina-Solexa 1G and preliminary ABI SOLiD) to reference genomes and derives genotype calls for diploid genomes.

Key Features:

• Mapping quality: Quantifies confidence that a read is correctly aligned to its designated position on a reference genome.
• Short-read alignment: Efficiently aligns tens-of-base-pairs reads typical of early Illumina-Solexa 1G data and provides preliminary handling of ABI SOLiD data.
• Mate-pair information: Leverages mate-pair relationships to improve alignment accuracy and disambiguate placements.
• Per-alignment error probabilities: Estimates error probabilities for each read alignment to enhance mapping reliability.
• Assembly by mapping: Builds assemblies by mapping shotgun short reads to a reference genome to derive consensus sequences.
• Genotype calling from quality scores: Uses base quality scores to derive genotype calls from the consensus sequence of a diploid genome.
• Bayesian statistical model: Applies a Bayesian model that incorporates mapping qualities and raw sequence quality score-derived error probabilities for genotype inference.
• Diploid sampling and correlated errors: Models sampling of two haplotypes and uses an empirical model for correlated errors at specific sites.
• Benchmarking: Was evaluated alongside mapping tools such as Bowtie, BWA, SOAP2, RMAP, GSNAP, Novoalign, and mrsFAST to characterize speed–accuracy trade-offs.

Scientific Applications:

• Short-read mapping: Aligns short reads to reference genomes for downstream variant and consensus analysis.
• Genotype calling in diploid genomes: Calls genotypes from mapped short reads using base quality scores and Bayesian inference.
• Reference-guided assembly: Constructs consensus sequences by mapping shotgun short reads to a reference genome.
• Method benchmarking: Serves as a reference in comparative evaluations against other read mappers to assess speed and accuracy trade-offs.

Methodology:

Implements mapping quality scoring, leverages mate-pair information, estimates per-read alignment error probabilities, maps shotgun short reads to a reference to build consensus, and performs genotype calling via a Bayesian model that integrates mapping qualities, raw sequence quality-derived error probabilities, sampling of two haplotypes, and an empirical model of correlated site-specific errors.