Genetic summary statistics available in GeLaTo

The current release of GeLaTo provides genetic summary statistics based on pairwise genetic distances. These are calculated as FST using the method introduced in Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure.

FST distances are used to infer rough population divergence times. Divergence time between two populations (as generations ago) is extrapolated from FST, being proportional to the effective population size (Ne) with a formula equivalent to Time = 2Ne * linearized FST (M. Nei, Molecular evolutionary genetics. Columbia University Press, 1987). Divergence time in year is calculated with a generation time of 29 years.

Effective population size Ne is calculated with IBDNe (S. R. R. Browning, B. L. L. Browning, Accurate Non-parametric Estimation of Recent Effective Population Size from Segments of Identity by Descent. Am. J. Hum. Genet. 97, 404–418 (2015)), which is based on Identity By Descent block coalescent. From the size and the number of Identity by Descent blocks it is possible to reconstruct the number of shared ancestors and infer variation in Ne through time. Identity by Descent blocks are retrieved after phasing the data with Beagle and running refinedIBD and its associate tools for gap removals (**B. L. Browning, S. R. Browning, Improving the accuracy and efficiency of identity-by-descent detection in population data. Genetics 194, 459–471 (2013).**). The harmonic mean of the last 50 generations was used to approximate Ne (this would minimize the effect of increase or decline in the last 10-20 generations). Populations with an Ne >10,000 were not considered (such high Ne can be resulting from population substructure and are not compatible with estimates coming from whole genome data). Ne values were considered for the following analysis only if the reconstructed profile was relatively stable in time, without large increase or decline (populations with a difference in Ne through generations of more than 10,000 were excluded). Populations with very large confidence intervals and continuous increase/decline were then further excluded. Divergence time estimates are available only for pairs of populations for which it is possible to calculate the Ne.