hapROH: Calling Runs of Homozygosity from low coverage ancient DNA
This method identifies genomic marks of parental relatedness, long runs of homozygosity (ROH). It can use low coverage human DNA data as low as 0.3x coverage and tolerates contamination and genotyping errors. The inferred segements are useful for two main purposes: 1. Identifying highly consanguineous individuals whose parents were closely related parents (such as first or second cousins). 2. Estimating more distant, background relatedness, which is informative about effective population sizes.
The program is designed to work out of the box for the data type most commonly used in ancient DNA, pseudo-haploid eigenstrat files. I maintain a Python software package, [hapROH], that can be installed via the Python package index (pip). A detailed description is available at the software page. In addition to the ROH calling program, a diverse set of functions to analyze and visualize the output are bundled with this package. Hands-on vignette notebooks walking you through the method and visualizations are available here. Feedback and requests for features are welcome!
Inferring Dispersal from long Identity by Descent Tracts
In my PhD, I developed a likelihood framework to fit geographic patterns of Identity by Descent tracts, the genomic marks of distant cousins. It fits the strong isolation by distance in population occupying a 2D habitat. Python classes implementing this inference scheme are available at my github page.