Optimized & Highly Heterozygous SNP Reference VCF for FACETS (hg38/GRCh38)

1. Summary

This deposit contains an optimized reference VCF file for the human genome (hg38/GRCh38). It lists common Single Nucleotide Polymorphisms (SNPs) specifically filtered for high population heterozygosity.

This file is designed for use with allele-specific copy number analysis tools, particularly FACETS. It replaces previous methods relying on fixed-distance thinning with a biologically relevant filtering approach based on 1000Genomes population frequencies.

2. Rationale

Standard dbSNP files present significant challenges for somatic CNV/LOH analysis:

Excessive Noise: They contain millions of rare variants or homozygous sites that provide no information for allelic imbalance detection.

Computational Overhead: Processing uninformative sites significantly slows down snp-pileup.

Biological Irrelevance: Chromosomes Y and M do not follow the diploid segregation models used by standard CNV algorithms.

This optimized reference solves these issues by retaining only common, strictly biallelic SNPs with a Reference Allele Frequency (RAF) between 0.25 and 0.75. This ensures that the chosen markers have the highest probability of being heterozygous in a patient, maximizing the resolution of FACETS while drastically reducing file size.

3. Contents of the Deposit

facets_reference_snps_hg38_optimized.chr_style.vcf.gz

The primary SNP reference file (UCSC style chr1, chr2, etc.).

facets_reference_snps_hg38_optimized.chr_style.vcf.gz.tbi

The Tabix index for the above file.

facets_reference_snps_hg38_optimized.no_chr_style.vcf.gz

The primary SNP reference file (style 1, 2, etc.).

facets_reference_snps_hg38_optimized.no_chr_style.vcf.gz.tbi

The Tabix index for the above file.

Galaxy-Workflow-Optimed_VCF_for_facets.ga

The .ga file allows users to import and reproduce the exact generation pipeline in a Galaxy instance.

Zenodo_fig1.png

Figure 1: Distribution of inter-SNP distances per chromosome. It demonstrates that despite filtering for heterozygosity, the average distance remains close to ~1kb (Avg: ~900-1100 bp), maintaining excellent coverage resolution.

Zenodo_fig2.png

Figure 2: Distribution of Reference SNP frequencies. This validates the filtering step, showing a clean cut-off at 0.25 and 0.75 RAF.

4. Generation Workflow (Transparency)

The generation process has been fully automated using a Galaxy Workflow to ensure reproducibility. The legacy Python script has been replaced by this workflow.

Methodology:

The reference VCF file (facets_reference_snps_hg38_optimized.chr_style.vcf.gz) was generated from the full NCBI dbSNP GRCh38 database (Build 157, GCF_000001405.40.gz) using a custom Galaxy workflow. The processing pipeline involved the following steps to ensure compatibility with snp-pileup and optimize for FACETS CNV analysis:

Initial Filtering: The raw VCF was filtered using bcftools view to retain only variants annotated as common (INFO/COMMON=1) and strictly classified as SNPs (TYPE=“snp”), excluding indels and rare variants.

Chromosome Renaming: Chromosome identifiers were mapped from NCBI RefSeq format (e.g., NC_000001.11) to UCSC style (e.g., chr1) using bcftools annotate. The mapping table was derived dynamically from the NCBI Assembly Report (GCF_000001405.40_GRCh38.p14_assembly_report.txt).

Frequency Normalization: The original INFO/FREQ field, containing text-based frequencies from multiple studies, was parsed using regular expressions. Specifically:

Only frequency data from 1000Genomes was retained.

The field was simplified from a complex string to a single Float value representing the Reference Allele Frequency.

The VCF header was reconstructed to redefine INFO/FREQ as Type=Float to allow for numerical filtering.

Contig Cleaning: To ensure biological relevance for somatic CNV analysis and reduce computational noise:

The Mitochondrial chromosome (chrM) and Y chromosome (chrY) were removed.

Non-canonical contigs (patches, decoys, and random contigs) were excluded from both the variant entries and the VCF header.

Heterozygosity Optimization: A final filter was applied using bcftools view to retain only SNPs with a 1000Genomes Reference Allele Frequency between 0.25 and 0.75. This step selects for high population heterozygosity, maximizing the utility of the markers for tracking allelic imbalance.

Final Output: The resulting VCF was reheaded and compressed using bgzip to create a standard, indexed VCF.gz file.

5. Recommended Usage

Download facets_reference_snps_hg38_optimized.chr_style.vcf.gz and its index.

Provide this VCF file as the SNP reference to the snp-pileup tool.

Example command:

snp-pileup -q15 -Q20 facets_reference_snps_hg38_optimized.chr_style.vcf.gz normal.bam tumor.bam | gzip > pileup.csv.gz

6. Authors and Citation

This resource was prepared by Christophe Antoniewski. Source data is derived from NCBI dbSNP 157. If you use these files in your work, please cite this Zenodo deposit and its doi.