Today Gary Olsen, who was my Ph.D advisor at the University of Illinois, gave a talk here in town at SDSU. The listed title was "Lateral Gene Transfer in Prokaryotes", which I was going to kid him about as Gary is a former postdoc of Norm Pace (who famously objects to the term "prokaryote"). But it turns out that the title was actually supplied by his host, so no dice there.
The talk dealt with further work in the Olsen lab on the topic of codon bias in relation to lateral gene transfer -- a topic which dates back to my time in the lab in the 1990s -- in fact Gary presented a figure from my dissertation near the beginning of the talk -- a PCA projection of E. coli codon usage showing the "rabbit head" pattern initially proposed by Médigue. In this projection, codon usage is shown to form three clusters, a cluster of "normal" genes (the "head"), an "ear" of known highly expressed genes, and an "ear" of genes of unusual codon usage containing many genes (such as integrases) thought to be recently transferred into the genome. The idea is that over time, transferred genes will assimilate into the codon usage of the host, much as immigrants lose their accents, making them impossible to detect through codon usage.
The obvious question is whether we can we use the codon usage of recent arrivals to determine their origin. In my day, the answer was no, we couldn't -- but we only had about 20 genomes available and so it was plausible that nothing close to the source organism had been sequenced. Now with over 1,000 genomes at our disposal, can we finally answer this question? Not exactly, but Gary and colleagues have discovered an interesting property -- the presumed transferred genes in E. coli and relatives seem to have a closer codon usage to *each other* regardless of host rather than being similar to any known organism, suggesting that they form a collection of genes that hop around from genome to genome, never staying around long enough to assimilate into the host -- it isn't a case of being transferred from the core of genome X to the fringe of genome Y as we supposed. They have found similar collections for other phylogenetic groups as well. This is really quite a different way to think about horizontally transferred genes.