---
license: apache-2.0
tags:
- biology
- genomics
- dna
- variant-effect-prediction
configs:
- config_name: full
  data_files:
  - split: validation
    path: full/validation.parquet
  - split: test
    path: full/test.parquet
- config_name: 10k_intergenic_variant
  data_files:
  - split: validation
    path: 10k_intergenic_variant/validation.parquet
  - split: test
    path: 10k_intergenic_variant/test.parquet
- config_name: 10k_intron_variant
  data_files:
  - split: validation
    path: 10k_intron_variant/validation.parquet
  - split: test
    path: 10k_intron_variant/test.parquet
- config_name: 10k_upstream_gene_variant
  data_files:
  - split: validation
    path: 10k_upstream_gene_variant/validation.parquet
  - split: test
    path: 10k_upstream_gene_variant/test.parquet
- config_name: 10k_downstream_gene_variant
  data_files:
  - split: validation
    path: 10k_downstream_gene_variant/validation.parquet
  - split: test
    path: 10k_downstream_gene_variant/test.parquet
- config_name: 10k_3_prime_UTR_variant
  data_files:
  - split: validation
    path: 10k_3_prime_UTR_variant/validation.parquet
  - split: test
    path: 10k_3_prime_UTR_variant/test.parquet
- config_name: 10k_synonymous_variant
  data_files:
  - split: validation
    path: 10k_synonymous_variant/validation.parquet
  - split: test
    path: 10k_synonymous_variant/test.parquet
- config_name: 10k_missense_variant
  data_files:
  - split: validation
    path: 10k_missense_variant/validation.parquet
  - split: test
    path: 10k_missense_variant/test.parquet
- config_name: 10k_5_prime_UTR_variant
  data_files:
  - split: validation
    path: 10k_5_prime_UTR_variant/validation.parquet
  - split: test
    path: 10k_5_prime_UTR_variant/test.parquet
- config_name: 10k_grouped_splice_region
  data_files:
  - split: validation
    path: 10k_grouped_splice_region/validation.parquet
  - split: test
    path: 10k_grouped_splice_region/test.parquet
- config_name: 10k_non_coding_transcript_exon_variant
  data_files:
  - split: validation
    path: 10k_non_coding_transcript_exon_variant/validation.parquet
  - split: test
    path: 10k_non_coding_transcript_exon_variant/test.parquet
- config_name: 10k_grouped_start_stop
  data_files:
  - split: validation
    path: 10k_grouped_start_stop/validation.parquet
  - split: test
    path: 10k_grouped_start_stop/test.parquet
- config_name: 10k_all_consequences
  data_files:
  - split: validation
    path: 10k_all_consequences/validation.parquet
  - split: test
    path: 10k_all_consequences/test.parquet
  default: true
- config_name: 20k_intergenic_variant
  data_files:
  - split: validation
    path: 20k_intergenic_variant/validation.parquet
  - split: test
    path: 20k_intergenic_variant/test.parquet
- config_name: 20k_intron_variant
  data_files:
  - split: validation
    path: 20k_intron_variant/validation.parquet
  - split: test
    path: 20k_intron_variant/test.parquet
- config_name: 20k_upstream_gene_variant
  data_files:
  - split: validation
    path: 20k_upstream_gene_variant/validation.parquet
  - split: test
    path: 20k_upstream_gene_variant/test.parquet
- config_name: 20k_downstream_gene_variant
  data_files:
  - split: validation
    path: 20k_downstream_gene_variant/validation.parquet
  - split: test
    path: 20k_downstream_gene_variant/test.parquet
- config_name: 20k_3_prime_UTR_variant
  data_files:
  - split: validation
    path: 20k_3_prime_UTR_variant/validation.parquet
  - split: test
    path: 20k_3_prime_UTR_variant/test.parquet
- config_name: 20k_synonymous_variant
  data_files:
  - split: validation
    path: 20k_synonymous_variant/validation.parquet
  - split: test
    path: 20k_synonymous_variant/test.parquet
- config_name: 20k_missense_variant
  data_files:
  - split: validation
    path: 20k_missense_variant/validation.parquet
  - split: test
    path: 20k_missense_variant/test.parquet
- config_name: 20k_5_prime_UTR_variant
  data_files:
  - split: validation
    path: 20k_5_prime_UTR_variant/validation.parquet
  - split: test
    path: 20k_5_prime_UTR_variant/test.parquet
- config_name: 20k_grouped_splice_region
  data_files:
  - split: validation
    path: 20k_grouped_splice_region/validation.parquet
  - split: test
    path: 20k_grouped_splice_region/test.parquet
- config_name: 20k_non_coding_transcript_exon_variant
  data_files:
  - split: validation
    path: 20k_non_coding_transcript_exon_variant/validation.parquet
  - split: test
    path: 20k_non_coding_transcript_exon_variant/test.parquet
- config_name: 20k_grouped_start_stop
  data_files:
  - split: validation
    path: 20k_grouped_start_stop/validation.parquet
  - split: test
    path: 20k_grouped_start_stop/test.parquet
- config_name: 20k_all_consequences
  data_files:
  - split: validation
    path: 20k_all_consequences/validation.parquet
  - split: test
    path: 20k_all_consequences/test.parquet
---

# Maize Allele Frequency Dataset

Source data described in [PlantCaduceus paper](https://www.pnas.org/doi/10.1073/pnas.2421738122). Raw data available at [plantcad/maize-allele-frequency-raw-data](https://huggingface.co/datasets/plantcad/maize-allele-frequency-raw-data/).

## Splits

- **validation**: Chromosomes 1, 3, 5, 7, 9
- **test**: Chromosomes 2, 4, 6, 8, 10

## Configurations

### full
Complete dataset with all variants.

### Subsampled Configurations
Subsampled datasets (excluding repeats) with max N variants, filtered by consequence type:
- `{max_n}_all_consequences`: Balanced sample across all consequence types (~N/11 per consequence)
- `{max_n}_{consequence}`: Up to N variants of specific consequence type

Available sample sizes: 10k, 20k

Available consequence types: intergenic_variant, intron_variant, upstream_gene_variant, downstream_gene_variant, 3_prime_UTR_variant, synonymous_variant, missense_variant, 5_prime_UTR_variant, grouped_splice_region, non_coding_transcript_exon_variant, grouped_start_stop

## Consequence Types

Individual consequence types and their grouped variants:
- **grouped_splice_region**: splice_polypyrimidine_tract_variant, splice_region_variant, splice_donor_region_variant, splice_donor_variant, splice_donor_5th_base_variant, splice_acceptor_variant
- **grouped_start_stop**: stop_gained, stop_lost, start_lost

Filtered consequences (excluded from dataset): stop_retained_variant

## Data Schema

- `chrom`: Chromosome
- `pos`: Position
- `ref`: Reference allele
- `alt`: Alternate allele
- `AC`: Allele count
- `AN`: Allele number
- `AF`: Allele frequency
- `MAF`: Minor allele frequency
- `consequence`: Consequence type (after grouping)
- `original_consequence`: Original Ensembl VEP consequence (before grouping)