Pipeline Overview¶

VNtyper 2 implements a multi-stage pipeline for genotyping MUC1 coding Variable Number Tandem Repeat (VNTR) variants associated with Autosomal Dominant Tubulointerstitial Kidney Disease (ADTKD-MUC1). The pipeline accepts BAM, CRAM, or paired-end FASTQ input and produces a genotyping result with confidence-annotated variant calls.

Pipeline Architecture¶

flowchart TD
    A[BAM / CRAM / FASTQ] --> B[Input Processing]
    B --> C{SHARK enabled?}
    C -->|Yes| D[SHARK Filtering]
    C -->|No| E[fastp QC]
    D --> E
    E --> F[BWA Alignment]
    F --> G[Coverage Calculation]
    G --> H[Kestrel Genotyping]
    H --> I[Postprocessing]
    I --> J[Scoring & Confidence]
    J --> K[Flagging]
    K --> L[Variant Selection]
    G --> M{adVNTR enabled?}
    M -->|Yes| N[adVNTR Genotyping]
    N --> O[Cross-Match]
    L --> O
    O --> P[Report Generation]
    L --> P
    M -->|No| P

Pipeline Stages¶

Input Processing¶

Handles BAM/CRAM region extraction (MUC1 locus), FASTQ quality control via fastp, unmapped read recovery, and coverage calculation over the VNTR region. Detects reference assembly and alignment pipeline from BAM headers.

Kestrel Genotyping¶

The core genotyping engine. Kestrel performs mapping-free, k-mer-based variant calling against the MUC1 VNTR reference. The postprocessing pipeline filters, scores, and annotates variants through nine distinct steps. This is the most critical component of VNtyper 2.

Scoring and Confidence Assignment¶

Calculates frame scores to identify frameshift mutations, computes depth-based confidence scores, and assigns precision labels (High_Precision*, High_Precision, Low_Precision, Negative) using empirically derived thresholds from Saei et al. (2023).

Flagging¶

Applies configurable post-hoc empirical filters to flag potential false positives and duplicate variants. Flags are evaluated before variant selection so that unflagged variants are preferred.

Optional Modules¶

Two optional modules provide complementary analyses: adVNTR (profile-HMM genotyping for independent validation) and SHARK (rapid MUC1 read extraction from large FASTQ datasets). Cross-matching logic compares Kestrel and adVNTR calls.

Report Generation¶

Produces an HTML report with variant summary tables, embedded IGV genome browser views, QC metrics, and screening interpretation. Cohort-level reports aggregate results across multiple samples with interactive Plotly charts.

Reference¶

Saei H. et al., iScience 26, 107171 (2023). DOI: 10.1016/j.isci.2023.107171