MERIT

MERIT quantifies erroneous substitutions and small insertions and deletions (indels) in ultra-deep high-throughput sequencing data to characterize context-specific sequencing error rates for improved mutation calling.

Key Features:

• Error Quantification: Quantifies erroneous substitutions across 96 possible nucleotide substitutions and four single-base plus sixteen double-base indels.
• Genomic Context Consideration: Analyzes nucleotides at the immediate 5' and 3' positions and trinucleotide contexts to assess local sequence effects on error rates.
• Error Rate Variation Analysis: Measures context-dependent variation in error frequencies in ultra-deep data (up to 1,300,000×), for example higher T>G transversions in GTC versus ATA and higher C>T transitions in GCG versus TCT.
• Depth Optimization: Implements an in silico approach to evaluate how sequencing depth (noting performance beyond 500× up to 1,300,000×) affects error rate plateauing and optimal depth selection to minimize errors relative to true mutations.
• Variant Calling Approach: Applies an all-inclusive variant calling strategy to quantify erroneous substitutions and small indels.
• Clinical Implications: Profiles specimen- and assay-specific sequencing artifacts to reduce false positives and negatives in mutation calling for clinical research on heterogeneous samples.

Scientific Applications:

• Ultra-deep sequencing error profiling: Characterizes sequencing artifacts in ultra-deep datasets (exceeding 500× and up to 1,300,000×) to map context-specific error landscapes.
• Mutation detection in clinical research: Differentiates true mutations from artifacts in heterogeneous clinical samples to improve accuracy of diagnostic and therapeutic decision-making.
• Sequencing assay optimization: Informs optimal sequencing depth and assay design by revealing context-specific error behaviors that impact variant calling sensitivity and specificity.

Methodology:

Applies an all-inclusive variant calling approach to quantify error rates across 96 substitutions and defined indel classes, analyzes immediate 5' and 3' bases and trinucleotide contexts, and performs in silico sequencing-depth optimization on ultra-deep datasets (up to 1,300,000×).