In this project I aim to study a group of ancient hunter-gatherers living in South Africa 2000-10000 years ago. The focus will be on diploid analyses where these individuals genomic characteristics will be compared with modern individuals from the 1000 genome project as well as with a couple of modern day Khoe-San populations living in the nearby area. The hypothesis is that this ancient group of people are the ancestors of modern day Khoe-San populations, but how has different demographic events and possible changes in selection pressure shaped the Khoe-San people through time? We have now expanded the project with trying to detect sapiens specific selective sweeps bt leveraging deep divergence and machine learning. This will be done by applying a random forest model to sapiens-specific population branch statistics (sPBS) and explore boundaries for detecting selective sweeps in terms of its strength and timing