PCA

Genotype likelihood approach

Matteo Fumagallis methods for doing PCA/Covariance based on ANGSD output files:

Fumagalli, M, Vieira, FG, Korneliussen, TS, Linderoth, T, Huerta-Sánchez, E, Albrechtsen, A, Nielsen, R (2013). Quantifying population genetic differentiation from next-generation sequencing data. Genetics, 195, 3:979-92.

This works without the need to call SNPs or genotypes based on genotype likelihoods.

NB! If you have very different depths for the different samples, e.i. some very low and others medium and high, then you might want to use the single base sampling approach PCA_MDS

The main documentation for this is found here: https://github.com/mfumagalli/ngsTools and here https://github.com/mfumagalli/ngsTools#ngscovar

single read sampling

Both PCA and MDS can be performed based on sampling of a single read at each site. This can work even with very low depth data e.g. <1X. This method can be found here:PCA_MDS. However, it requires low error rate and polymorphic sites need to be inferred (or provided by user)