HiMWA

A Hierarchical Multiple-Wave Admixture Model for Inferring Complex Admixture Histories

Please ensure HiMWA.py is in the same directory as the src/ folder.

Version

The analyses presented in the manuscript were performed using HiMWA v1.0.0, available at: https://github.com/Shuhua-Group/HiMWA/releases/tag/v1.0.0

Input Files

Ancestral tracts can be obtained from standard local ancestry inference (LAI) tools such as RFMix2 or FLARE. To facilitate reproducibility and broader application, we provide example scripts (in the other_script/ directory, flare2seg.py and msp2seg.py) that convert LAI outputs into HiMWA-compatible segment files.
The format of input ancestral tracts file is same as HierarchyMix. Each row represents an ancestral tract characterized by the following parameters: (1) the genetic distance of start-point (in Morgans), (2) the genetic distance of end-point (in Morgans), (3) the ancestry of origin, (4) the index of admixed haplotype, and (5) the chromosome label. All genetic positions are specified in Morgans (M) with decimal precision. An example segmentation file example.seg is provided in the examples/ directory.

To run with this example data:

python HiMWA.py --input examples/example.seg --output examples/example

Optional arguments allow more refined analyses. For example, bootstrap resampling and tract-length filtering can be specified as:

python HiMWA.py --input examples/example.seg --output examples/example --bootstrap 100 --lower 0.003 --lowerEF 0.02

Output Files

The format of output model file is same as MultiWaver series and HierarchyMix software. The example generates an output file example.txt.

Arguments and Options

Flag	Type	Description	Default
--input	string	Input file name (required)	-
--output	string	Output file prefix	out
--lower	float	Min tract length cutoff	0
--lowerEF	float	Min tract length (admixed ancestors)	0
--bootstrap	int	Bootstrap iterations	0
--ci	string	Confidence level	0.95
-h, --help	flag	Show help	-

Notes

It should be noted that the model selection of HiMWA may be unreliable for admixed populations with extremely biasd admixture proportions. Furthermore, in scenarios with multiple recent admixture waves, the inferred times of recent admixture events by HiMWA may be overestimated. It is advised to correct the overestimation of the admixture times referring the strategy in this study under specific model configurations.