Are individual and population IDs consistent across treeseq files?

[clawsoon]

Do the treeseq files at https://zenodo.org/record/5495535 all have consistent individual and population IDs? Or is it safer to use metadata names + sources to get consistent identities?

[jeromekelleher]

Good question! @awohns I guess we should document the metadata and samples that are included somewhere and whether these IDs are stable across the files?

[awohns]

Thanks for the question @clawsoon. @jeromekelleher, metadata for individuals and populations is documented in the description at https://zenodo.org/record/5495535. These IDs are stable across the files.

[clawsoon]

Hi @awohns. Are you saying that only the metadata IDs are stable across the files? Or are the IDs within the treeseq files also consistent (e.g. those used by ts.individual(i) and ts.population(i) or returned by variant.genotypes())? Thanks.

[clawsoon]

FWIW, this is what I've been using to get unique identities for individuals and populations. Not sure if it's the best way.

    for individual in ts.individuals():
        individual_metadata = json.loads(individual.metadata)
        if 'sgdp_id' in individual_metadata:
            source = 'SGDP'
            individual_name = individual_metadata['sgdp_id']
        elif 'sample' in individual_metadata:
            source = 'HGDP'
            individual_name = individual_metadata['sample']
        elif 'individual_id' in individual_metadata:
            source = 'TGP'
            individual_name = individual_metadata['individual_id']

        population = ts.population(ts.node(individual.nodes[0]).population)
        population_metadata = json.loads(population.metadata)
        population_name = population_metadata['name']
        unique_population = (source, population_name)