UMI-tagged cfDNA sources

Ask for help
1

Hello, I need simulated UMI-tagged cfDNA files (forward and reverse) to test my pipeline. Having a ground truth would be a huge plus. Do you know where I can get such simulated data?

2

Consider asking this to Seqera’s AI (Seqera AI Chat). This isn’t my field, but perhaps you can evaluate whether this is useful to you.

Here’s it’s output:

Based on my research, I can provide you with several excellent options for simulated UMI-tagged cfDNA files with ground truth data for testing your pipeline:

:dna: Simulation Tools for UMI-tagged cfDNA Data

1. SafeMut - The Premier UMI-aware cfDNA Simulator

SafeMut is the first and most sophisticated UMI-aware NGS variant simulator specifically designed for cfDNA analysis:

  • GitHub: Available as open-source software
  • Key Features:
    • Specifically designed for cell-free DNA with UMIs
    • Incorporates allele-fraction overdispersion modeling
    • Generates realistic tumor/normal FASTQ pairs
    • Validated against FDA SEQC2 datasets
    • Ground truth variants with known VAF (variant allele frequency)
    • Forward and reverse read simulation
  • Input Requirements: BAM file + VCF file
  • Output: Paired tumor/normal FASTQ files with UMIs

2. Combination Approach: ART + BAMSurgeon

For more general simulation with cfDNA characteristics:

  • ART (Illumina mode): For realistic read simulation with cfDNA parameters
    • Mean insert size: 166bp (typical for cfDNA fragments)
    • Insert size standard deviation: 24
    • Read length: 100bp
    • Profile: HiSeqX TruSeq for realistic quality scores
  • BAMSurgeon: For spiking in variants at known frequencies
    • Can add variants at specific VAF levels (0.1% and higher)
    • Maintains realistic coverage patterns

3. Public Reference Datasets with Ground Truth

SEQC2 Consortium Datasets

  • Source: FDA-led Sequencing Quality Control Phase 2
  • Contains: Well-characterized cfDNA samples with known variants
  • Advantage: Rigorously produced with detailed protocols
  • Access: Publicly available for research use
  • Ground Truth: Includes validated variant calls and confident regions

SeraCare Reference Materials

  • Purpose: Commercial cfDNA reference standards
  • Features: Known mutation profiles with certified VAFs
  • Applications: Widely used for assay validation
  • Availability: Can be purchased for research use

:hammer_and_wrench: Recommended Workflow for Creating Test Data

Option 1: Using SafeMut (Recommended)

`# 1. Get a normal cfDNA BAM file (from SEQC2 or similar)

2. Create a VCF with variants you want to test

3. Run SafeMut to generate synthetic data

safemut -i normal.bam -v variants.vcf -o output_prefix`

Option 2: ART + BAMSurgeon Pipeline

`# 1. Generate baseline cfDNA reads with ART
art_illumina -i reference.fa -l 100 -f 30 -m 166 -s 24
–profile HSXt -o cfDNA_reads

2. Align to reference

3. Spike in variants with BAMSurgeon

bamsurgeon -i aligned.bam -v variants.vcf -o spiked.bam`

:bar_chart: Ground Truth Components You’ll Get

  1. Known Variants: VCF files with exact positions and allele frequencies
  2. UMI Information: Synthetic UMIs with realistic error patterns
  3. Fragment Characteristics: Realistic cfDNA fragment size distributions
  4. Coverage Patterns: Authentic sequencing depth variations
  5. Quality Scores: Technology-specific error profiles

:bullseye: Pipeline Testing Strategy

  1. Start Small: Use SafeMut with a few known variants at different VAFs
  2. Gradual Complexity: Add more variants and lower allele frequencies
  3. Validation: Compare your pipeline’s calls against ground truth
  4. Performance Metrics: Calculate sensitivity, specificity, and precision

Would you like me to help you implement any of these simulation approaches, or would you prefer guidance on accessing the SEQC2 datasets? I can also assist with setting up the specific tools mentioned above.

1 Like