Building 1000 Genomes Dataset¶
This step builds the canonical dataset used by the neural genotype predictor. It starts from 1000 Genomes metadata, phased VCFs, a reference genome, and AlphaGenome track generation.
Run the commands below in order. Use the links at the end of the page when you need conceptual or component-level details.
1. Inspect Data Configuration¶
Start with the default non-longevous dataset builder config. Confirm local paths such as metadata_csv, reference.fasta, vcf_pattern, and output settings before running a large job.
For larger or more specific builds, use one of the configs under:
The config points to 1000 Genomes metadata, reference FASTA, VCF patterns, sample selection rules, AlphaGenome output settings, and output directories.
2. Discover AlphaGenome Tracks¶
Before large runs, export track metadata so you know which output families and ontology terms are available:
genomics alphagenome tracks \
--api-key API_KEY \
--output configs/workflows/alphagenome/tracks.json
For sequence sizing and output selection, see AlphaGenome Sequences And Outputs.
3. Build The Dataset¶
Run the builder:
genomics dataset-builders non-longevous build \
--config configs/workflows/non_longevous_dataset/default.yaml
The resulting canonical layout contains dataset metadata, reference windows, per-individual metadata, haplotype sequences, and AlphaGenome prediction directories.
4. Validate Dataset Availability¶
When the dataset is registered locally, validate it with:
If you plan to train aligned DITA models, also check BCFtools chain artifacts:
genomics audit-data \
--dataset-id 1kg_high_coverage \
--check-bcftools-chain \
--sample-limit 3 \
--fail-on-missing
More Details¶
| Topic | Read More |
|---|---|
| Data roots and dataset IDs | Data And Results |
| Canonical dataset layout and audit checks | Data Registry |
| AlphaGenome workflow | AlphaGenome |
| Dataset builder internals | Dataset Builders |
| Configuration conventions | Configuration |