Usage: 

1. Calculate the co-ancestry matrix: ./RADpainter paint INPUT_RAD_FILE.txt

2. Assign individuals to populations: ./finestructure -x 100000 -y 100000 -z 1000 INPUT_RAD_FILE_chunks.out INPUT_RAD_FILE_chunks.mcmc.xml

3. Tree building: ./finestructure -m T -x 10000 INPUT_RAD_FILE_chunks.out INPUT_RAD_FILE_chunks.mcmc.xml INPUT_RAD_FILE_chunks.mcmcTree.xml

 

Input format: 
The input file (INPUT_RAD_FILE.txt) should be in one of the following formats:

1. Stacks  populations program output with the --radpainter option. This is the easiest option if you have your data in Stacks.

2. Stacks v1 export_sql.pl output (-a haplo -o tsv -F snps_l=1): Example

3. Tag Haplotype Matrix (for unmapped data): Example

4. Chromosome/scaffold + Tag Haplotype Matrix (for mapped data): Example

In all cases, rows correspond to RAD loci. Data columns correspond to individual haplotype calls; only sites that are variable in the dataset are needed. In diploid individuals, if the two alleles are the same, e.g. TGAT/TGAT, they can be collapsed to just TGAT. If the two alleles differ then both alleles need to be fully specified (e.g. TGAT/CGAT). Missing data is left blank; see the examples above. If you have polyploid individuals then the best way is to write out all the alleles; e.g. TGAT/TGAG/TGAG/TGAG could be the four haplotypes for a tetraploid individual. 

 

If you have your data in a VCF file, the RADpainter hapsFromVCF command may work; feel free to get in touch if it doesn't work for your specific VCF. 

​

If the populations --radpainter option is for some reason not optimal, the fineRADstructure package also includes a python script Stacks2fineRAD.py that can be used for conversion from the standard output of the program "populations" in Stacks (a file usually called "batch_1.haplotypes.tsv"). It has various filtering options. The script is also available here. An example command line to run this would be: python Stacks2fineRAD.py -i Stacks_output.tsv -n 10 -m 30. The -n option, specifies the maximum number of SNPs allowed at a locus (misaligned paralogs can lead to unusually many SNPs). Using the -m option, you discard individuals with unusually high levels of missing data (e.g. using option -m 30 we set the cutoff to 30%). 

Finally, there is an utility script from Edgardo M. Ortiz enabling conversion from the format used by the pyRAD and ipyrad toolkits https://github.com/edgardomortiz/fineRADstructure-tools.

 

Note: If you have a reference genome, the RAD loci should ideally be ordered according to genomic coordinates. If you have unmapped loci, we provide a script sampleLD.R that can reorder loci according to linkage disequilibrium (LD). If LD is strong and loci are not sorted, this could lead to overconfident clustering. Therefore, we recommend using the sampleLD.R script before running RADpainter to users with unmapped data who want to be extra careful to ensure they obtain a CONSERVATIVE upper bound on the number of statistically identifiable clusters. Example command line: Rscript sampleLD.R -s 1 -n 500 INPUT_RAD_FILE.txt INPUT_RAD_FILE_reordered.txt. This should do the job. If you want to understand the options, they are described in the R script. 

 

Note 2: The amount of missing data should not vary too much across individuals (remove outliers if necessary) or systematically between putative populations (test a few runs with filtering tags/individuals at various stringency levels; as a rule of thumb, I'd say that missingness over 50% per individual can cause problems).