README: Geolocation-by-genotype reveals systematic and unidirectional mass reorientation in a migratory songbird Within this folder are the files needed to recreate the analysis presented in the associated manuscript. They are provided for the purpose of reviewing the manuscripts findings, not for applying the same techniques to new datasets. More accessible, generalist code will be provided for this purpose at the point of publication and will be made available online. Step 1: SNP filtering This script takes as input an ANGSD-generated beagle file (included in this repository), and returns a dataframe containing information about the goodness-of-fit of geographic models of allele frequency through space. The presets used to create the beagle file are included in the methods section of manuscript. The code is fully annotated, such that anyone with a working understanding of R might understan it. Step 2: LOOCV and GBG This script takes either a) an ANGSD-generated beagle file or b) a pre-filtered beagle file (i.e. a beagle file where specific lines have already been selected). It then: 1) Makes generalised additive models (GAMs) of allele frequency through space.2) Clusters alleles based on how they vary through space.3) Performs leave-one-out cross-validation (LOOCV) and makes plots of model performance.4) Geolocates samples based on the models created in step 1. The prerequisite datasets for step 2 (and, implicitly, the output from step 1) are included in this folder. The file xcol_ref contains metadata from samples. The variable test_train contains either a T or F value. T means that the sample in question is a georeferenced Siberian sample. F means that the sample in question is a vagrant sample from Europe. The rest of the values in xcol_ref are implicit.