Do PCA to see whether similar samples cluster together

[jwschroeder3]

Grab every nth position's data, where n is between 1,000 and 100,000, and run PCA. See which samples cluster together.

Would be a nice QC step.