BIGapp Workshop Dataset Guide

Atlantic Giant pumpkin diversity panel

Documentation, downloads, checksums, and expected results for the simulated BIGapp workshop dataset.
Author

Alexander M. Sandercock

Published

December 15, 2025

Modified

July 13, 2026

Dataset at a glance

  • Status: fully simulated teaching data
  • Samples: 200
  • Markers: 1,200 before filtering; 1,000 at the workshop checkpoint
  • Files: genotype VCF, phenotype CSV, SNP-information CSV, and README

This workshop uses a fully simulated diversity panel of 200 Atlantic Giant pumpkin (Cucurbita maxima) samples. It was designed for teaching and must not be interpreted as observed biological or breeding-program data.

Direct downloads

File Records SHA-256 checksum
atlantic_giant_pumpkin_diversity.vcf.gz 1,200 variants, 200 samples badf7aee0e145e90e0d1adff380f0fe561c702ac4bf9e2c5304d2e4037edc7da
atlantic_giant_pumpkin_diversity_phenotypes.csv 200 data rows 2c916cb8731570d3574f3e0d676bf84eb9c25d22ea10a94a1291b44d7adb253a
atlantic_giant_pumpkin_diversity_snp_info.csv 1,200 data rows 3a9213fafe4f752cc3f3efc654af6cb89206020a9f458a1b53bf471bd63f8512
Dataset README Documentation 99ce5346bf23ffc3d224e729582ce04f91cc298eb87c6fe59803ec89b4e3da90

Checksums identify the copied workshop files. On macOS or Linux, verify one with:

shasum -a 256 atlantic_giant_pumpkin_diversity.vcf.gz

What the VCF contains

The compressed VCF is based on coordinates from the C. maxima HZAU telomere-to-telomere assembly. It has 20 chromosomes, 1,200 biallelic SNPs, and 200 sample columns. Sample genotype fields use GT:DP:AD:

  • GT records the allele call.
  • DP records total read depth for one sample at one SNP.
  • AD records reads supporting the reference and alternate alleles.

The simulated quality categories contain 1,000 good markers, 100 low-MAF markers, and 100 high-missingness markers. The poor-quality markers are distributed across all chromosomes.

What the phenotype CSV contains

Each of the 200 rows represents one simulated sample. Sample_ID joins this table to the VCF. The remaining fields describe state, region, breeding program, sequencing batch, plate and well, three ancestry proportions, and Fruit_Weight_lbs.

The samples represent three overlapping simulated regions: Eastern, Midwest, and Western. Their ancestry proportions create structure that can be explored in PCA and accounted for in association analysis.

What the SNP-information CSV contains

Each of the 1,200 rows describes one VCF marker. The file contains chromosome, position, SNP ID, alleles, quality category, QTL annotation, and simulated ancestral allele frequencies.

SNP_0240 is explicitly the simulated ground-truth causal variant for fruit weight. SNP_0238 and SNP_0239 are simulated tag markers in strong linkage disequilibrium with it. This ground truth is available only because the data are simulated. A significant marker in a real study should not be called causal without additional evidence.

Verified checks

The local copies were checked for:

  • 1,200 VCF variant records
  • 200 VCF sample names
  • 200 phenotype records with unique sample IDs
  • 1,200 SNP-information records with unique marker IDs
  • complete overlap between phenotype IDs and VCF sample IDs
  • complete overlap between SNP-information IDs and VCF marker IDs

With ploidy 2, minimum depth 10, MAF 0.05, and 50 percent SNP and sample missingness thresholds, the teaching checkpoint is:

Result Expected value
Retained variants 1,000
Retained samples 200
Missing genotype cells after filtering 9,447

The missing-cell total is important: genotype cells below depth 10 are first set to missing, then the SNP and sample missingness filters are evaluated. Changing the depth cutoff can therefore change both missingness and which rows or columns pass their thresholds.

Expected analysis patterns

  • PCA: three overlapping regional clusters, with Western samples more separated and gene flow between regions.
  • GWASpoly: a chromosome 4 peak involving SNP_0238, SNP_0239, and SNP_0240, followed by one simulated QTL region after LD-window filtering.
  • Genomic prediction: numerical results depend on random cross-validation splits and selected settings, so the workshop does not assign universal quality labels or promise a fixed correlation.

Reference coordinate citation

The simulated coordinates refer to the HZAU assembly described by Zeng Q, Wei M, Li S, et al. (2024), “Complete genome assembly provides insights into the centromere architecture of pumpkin (Cucurbita maxima),” Plant Communications, 5(9):100935. https://doi.org/10.1016/j.xplc.2024.100935

For BIGapp, cite Sandercock AM, Peel MD, Taniguti CH, et al. (2025), “BIGapp: A User-Friendly Genomic Tool Kit Identified Quantitative Trait Loci for Creeping Rootedness in Alfalfa (Medicago sativa L.),” The Plant Genome, 18:e70067. https://doi.org/10.1002/tpg2.70067

Use the data: Start with the required local setup, then continue to Module 1: Data quality control.

Back to top