PCA summary plots rely on fam file over phenotype

[rkwalters]

Plots generated by pcaer (e.g. MDS plots, summary plot of PC association to phenotype) use phenotypes from the individuals FID tag (cas, con, mis) rather than from the fam file. This means that any updates to the phenotype after initial QC (e.g. for alternative outcomes, updated phenotyping, etc) that are applied to the fam file will not be reflected in the PCA output.

This is particularly problematic for the summary pdf, since the associations between the original phenotype (indicated in the FID) and the PCs will be reported rather than association with the updated/alternative phenotype, which may be misleading when using the pcaer output to identify necessary PC covariates for GWAS with the updated phenotype.

Affected code is in pca_plot_5 with the creation of the $mds_ow_file and subsequent use of the CC column in R for plotting (lines 350-600).