Whenever data is submitted to the protein data bank, the software packages used to clean up the data are always disclosed (See:
http://www.rcsb.org/structure/2OF5 ). The worse the resolution of the imaging technique used to obtain the data, the more likely there is aberrant data. Hydrogen bonds would never actually be 1.4A away.
I only measure hydrophobic interactions 2.7A or closer. Those are the only ones which help understand how any why the protein folded the way it did. The closest hydrophic interactions are squished because of secondary structures surrounding them.
A straight line can be drawn though two secondary structures. Call the two secondary structures point A and point C. The hydrophobic interactions caught in the middle are point B. There can also be three secondary structures 120 degrees away from each other, with hydrophobic interactions squished in the middle.
With good resolution, I've never measured hydrophobic interactions less than 2.0A away from each other. They tend to be about 0.5A farther away from each other than hydrogen bonds. Even 2.0A is rare. Seldom are they closer than 2.3A.
