Principle Component Analysis

This method performs a principle component analysis (PCA) using ANGSD and ngsPopGen for PCA calculation. Please see NGSPopGen for full details on this method.

Basic Usage

To run this method, use the following command

angsd-wrapper PCA Principal_Component_Analysis_Config

where Principal_Component_Analysis_Config is the full path to the configuration file for the PCA.

Input files

All inputs should be specified in Principal_Component_Analysis_Config.

Common Variables

This method does make use of Common_Config, those that are used are listed below:

Variable	Function
SAMPLE_LIST GROUP_SAMPLES on dev	A list of samples to be used in calculations
PROJECT	Name given to all outputs in ANGSD-wrapper
SCRATCH	Place to store files, the full path is SCRATCH/PROJECT/PCA
REGIONS	Limit the scope of ANGSD-wrapper to certain regions
GT_LIKELIHOOD	Estimates genotype likelihoods
N_CORES	Number of cores to use, please do not set above the limits of your system
DO_MAJORMINOR	Estimate major/minor alleles
DO_GENO	Call genotypes and setup the output
DO_MAF	Calculate per-site frequencies
DO_POST	Calculate the posterior probability using per-site frequencies

Method-Specific Variables

This method has no method-specifc variables

Method Parameters

The parameters for this method can be tweaked as necessary, they have been set for optimal generalized function:

Parameter	Function
CALL	Call genotype from maximum probability
N_SITES	Set the maximum number of sites to use. If left undefined, wrapper uses all valid sites.

Output files

Naming Scheme	Contents
PROJECT_PCA.arg	Details of arguments
PROJECT_PCA.covar	Results of the principle component analysis
PROJECT_PCA.geno	Genotype calls
PROJECT_PCA.mafs.gz	Per-site frequencies
PROJECT_PCA.backup.graph.me.[timestamp]	ngsPopGen doesn't voluntarily overwrite existing data files, so ANGSD-Wrapper moves existing *.graph.me files to a unique backup within the same directory in order to prevent repeated executions.

Visualization

PROJECT_PCA.covar (renamed to PROJECT_PCA.graph.me during processing) can be visualized with the Shiny graphing interface. A web browser with a graphical user interface is required.

(Optional) Subgroups

To compare clusters of known subgroups in the samples, create a .clst clusters file that contains labels for each sample. The .clst file should be formatted as follows, with the CLUSTER value determining the color of the data point within the final plot. The header line should be included exactly as shown as well.

FID	IID	CLUSTER
[Sample Name]	1	[Group Name]
WBDC_348	1	wild
Morex_jh100	1	elite
...	...	...