Dataset DOI: 10.5061/dryad.0zpc867cx

Description of the data and file structure

Input files with information about 5,358,142 bison-cattle ancestry informative markers necessary for using AncestryHMM to perform local ancestry inference of bison genomes to test for cattle ancestry. Sites are on the ARS-UCD1.2 cattle genome coordinates. Marker distances assume a flat recombination rate.

Ancestry-informative markers were identified using four high-coverage steppe bison genomes and 20 modern cattle genomes from four breeds. We obtained sites with at least 90% frequency difference between these two panels, and filtered sites to be fixed in a large panel of cattle genomes, be confidently called heterozygous ancestry in an F1 bison-cattle hybrid, and not have any reads carrying the cattle allele in any bison older than 1,000 years.

These three files contain the sites, alleles, and counts of each allele within each panel for each of the ancestry-informative markers. Files are compressed (gzipped) text files.

Files and variables

File: ancient_bison_cattle.combined.merge.freq_0.9.d0.min_ind_2_6.no_1k_bull.map_filt.no_del.buzz_filt.no_cattle_min11_anc.positions.txt.gz

Description: chromosome and position of inferred ancestry informative markers.

File: ancient_bison_cattle.combined.merge.freq_0.9.d0.min_ind_2_6.no_1k_bull.map_filt.no_del.buzz_filt.no_cattle_min11_anc.sites.txt.gz

Description: positions and alleles of ancestry-informative markers.

File: ancient_bison_cattle.combined.merge.freq_0.9.d0.min_ind_2_6.no_1k_bull.map_filt.no_del.buzz_filt.no_cattle_min11_anc.update_dist.ahmm.gz

Description: panel information for running AncestryHMM. Seven columns are present:

• chromosome
• position
• bison panel allele 1 count
• bison panel allele 2 count 
• cattle panel allele 1 count
• cattle panel allele 2 count
• distance since last marker

Code/software

Panel information is needed to run local ancestry inference (LAI) using Ancestry_HMM (https://github.com/russcd/Ancestry_HMM). Panels were constructed from genotype calls using ancient steppe bison and modern cattle. Panel allele counts can be combined with sample read or genotype counts to perform LAI. An example of how to do so is provided in run_ahmm.sh which takes a list of samples aligned to the ARS-UCD1.2 cattle genome and generates per-sample read counts at AIM sites using samtools (https://www.htslib.org) mpileup, merges with source panels, performs single-sample LAI using AncestryHMM, and summarizes the resulting ancestry posterior distribution.

The run_ahmm.sh example references a few helper scripts:

pu2ahmm_ssmode.py: generates ahmm input files from single-sample read pileups. For ancient samples that have been generated using single-strand library preparation methods, setting --single allows for strand-specific filtering of potentially damaged reads at transition sites.

posterior2tracts.v3.py: converts LAI posterior to ancestry tracts. With  --smooth adjacent tracts of the same ancestry are aggregated.

posterior_summary.py: summarizes overall posterior ancestry distribution.

posterior2geno.py: generates hard genotype calls from the posterior, using a posterior threshold.

ahmm_jackknife.1Mb_blocks.R: performs jackknifing to estimate the overall proportion of inferred cattle ancestry.

Access information

Other publicly accessible locations of the data:

• A preprint describing this work is available at https://www.biorxiv.org/content/10.64898/2025.12.24.696034v1
• Data and scripts can also be found at Bison_cattle_local_ancestry (https://github.com/jooppenh/Bison_cattle_local_ancestry)

Data were derived from the following sources:

• Sequencing data used to generate ancestry-informative markers is available from NCBI BioProjects PRJNA1374439