I looked over the paper pretty quickly, but on a quick read it looks really nice. The data are really impressive, and they seemed to have pulled out really interesting trends for how transcription factors are affecting coding sequence. Furthermore, they map how variants found across the various cell lines they used (81!) to show that these variants actually are causal for transcription factor binding changes. For those saying this is not a big deal, or is obvious; it is, and is not (respectively).
It's been pretty clear for some time that codon choice is not random (hence codon optimization is useful when taking a gene from one organism to another), and over the last few years it's also been clear that codon bias can be evolutionarily constrained. For a while, it mostly thought to be based on tRNA levels (basically anticodons). However, it's been increasingly clear that there are other constraints to protein coding sequences than just the decoded protein sequence (aka the genetic code).
For example, we published a paper a few weeks back [1] showing that in bacteria (and probably higher organisms), the N-terminus of genes has a lot of rare codons and this is due to other constraints such as relieving mRNA structure to allow better translation of proteins. I think in the coming years, we will find that other regulatory elements also shape this code, including sequences that control splicing, small RNAs, mRNA degradation and transport, et cetera
Anyways, it's a pretty fun time in biology. The tools we have now make studies that were ridiculously impossible just a few years ago, a reality for an individual lab. I can't wait to see what the next few years bring.
[1] http://arep.med.harvard.edu/pdf/Goodman_Sci_13.pdf\
EDIT: Since my comment has hijacked the most useful comment linking to the original study, I'll link to the comment here:
https://news.ycombinator.com/item?id=6897122