​Methods for Human Genomics

Available statistical genetic software packages were typically designed for urban cohorts in the US or western Europe, generally for populations of well mixed European ancestry. Recent advances in computational population genetics include algorithms for local ancestry assignment within ‘admixed’ individuals, but even these tend to be tailored to a narrow range of African-Americans or Latinx populations from 1000 Genomes. Inference of pedigree relationships, local ancestry haplotypes, GWAS, identity-by-descent etc. algorithms simply do not perform well in many global cohorts. Before we apply most off-the-shelf population genetic algorithms, we have to re-calibrate error rates or develop our own algorithms. Our current methods in development include: efforts to infer pedigree relationships in endogamous populations, create an improved long-read genome for Khoe-San descent individuals, identify biases in methylation probes that affect African ancestry individuals, and predicting environmental barriers to population structure.

Our software page introduces software packages we have designed.