Sequence deeper without sequencing more: Bayesian resolution of ambiguously mapped reads.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Next-generation sequencing (NGS) is crucial for genomic insights but struggles with repetitive sequences, leading to 15-30% of reads being unmappable.
• Discarding ambiguously mapped reads can cause data distortion, limiting comprehensive genomic analysis.
• Existing methods for handling ambiguous reads are often computationally intensive or have limited applicability.

Purpose of the Study:

• To develop a computationally efficient algorithm, SmartMap, that leverages ambiguously mapped reads in NGS data.
• To improve the accuracy and completeness of genomic analyses by incorporating reads that traditionally are discarded.
• To enable the study of repetitive genomic elements and their regulatory roles.

Main Methods:

• SmartMap augments standard aligners by assigning weights to ambiguously mapped reads using Bayesian analysis of read distribution and alignment quality.
• The algorithm is computationally efficient, processing billions of alignments in about an hour on a standard PC.
• Applied to peak-type NGS data (MNase-seq, ChIP-seq, ATAC-seq) across three organisms.

Main Results:

• SmartMap increased read depth by up to 53% and the mapped genome proportion by up to 18% compared to using only uniquely mapped reads.
• Enabled analysis of over 140,000 repetitive elements previously inaccessible to traditional ChIP-seq workflows.
• Provided new insights into the epigenetic regulation of repetitive elements.

Conclusions:

• Discarding ambiguously mapped reads poses significant risks of data distortion and missed biological discoveries.
• SmartMap offers an efficient and effective solution for utilizing ambiguously mapped reads, enhancing genomic analysis capabilities.
• The method unlocks new avenues for biological discovery, particularly in understanding the regulation of repetitive genomic regions.