paper about spread of b/f .. I think

   http://scienceblogs.com/effectmeasure/2007/03/a_paper_about_spread_of_bird_f.php



     Trouble at another NIH institute | Main

A paper about spread of bird flu. I think.

Posted on: March 7, 2007 7:33 AM, by revere

Let me apologize in advance. This is a bit of a rant about scientific writing. It didn't start out that way, but as I hit the keyboard, Satan took control.

A new paper in the Proceedings of the National Academy of Sciences (PNAS, or "****" in the trade) is said to be reporting that bird flu comes from southern China (Wallace et al., "A statistical phylogeogrpahy of influenza A H5N1," PNAS, March 13, 2007, 104:4473-4478). We already knew that. So what's new? That's a bit harder to say.

Here's the lede (i.e., the opening lines of a news story) from the Agence France Presse news agency:

US researchers have reconstructed the evolution of avian flu and its spread over the past decade from its first origins in southern China, according to a new study.
The team from Irvine University in California combined genetic and geographic data for the H5N1 virus, identifying many of the migration routes through which the strains spread across Asia and then around the globe. (AFP)


This sounds like they used genetic sequence information to reconstruct (i.e., construct again, from individual pieces of genetic and geographic information) the original migratory routes via birds. This is pretty interesting, so I wanted to know how they did it and how the reporter got this information from what turns out to be a highly technical paper. The researchers are identified as coming from the University of California at Irvine, so I went to the media office at UCI's website and found a press release there. It starts out this way:

UC Irvine researchers have combined genetic and geographic data of the H5N1 avian flu virus to reconstruct its history over the past decade. They found that multiple strains of the virus originated in the Chinese province of Guangdong, and they identified many of the migration routes through which the strains spread regionally and internationally. (UCI Press Release)
So this news article, like so many, is from a university media office press release. Not really a problem. It's a good way to find interesting stories and usually accurate as the releases come from information provided by the scientists themselves and usually checked by them. So that mystery is solved. But others appeared. The paper is not just highly technical but terse to the point of impenetrability, with much data and back-up information relegated to Supplementary files online, files to which I didn't have access (the paper has yet to appear on the PNAS website; I had a proof copy). But from what I read, there is no actual "reconstruction" of migratory pathways here but instead an ambiguous set of inferences producing a depiction, via a cartoon map, of how H5N1 might have spread from a few places to a few other places. There were no viral isolates from Europe, the Middle East or Africa included, so statements are restricted to China and nearby parts of Asia, Vietnam and Thailand, Indonesia and northern Russia.

I will admit I don't fully understand all the details of their method, which seems new and is not explained to any extent. But I'm not willing to take that as my failing, but theirs. They could have explained it so that a reasonably equipped flu person might understand it. PNAS is a general science publication, not a specialist one. And some of the parts I did decipher didn't give me much confidence communication was a high priority. In fact it felt as if technical jargon was being used as much to obscure things as reveal them. For example, the authors state that they "calculated the arc cosine of the scalar product of the migration matrices." The arc cosine of the scalar product between two vectors is just the angle between them. Why not just say that? It makes me think clarity was a low priority.

Here's another example. The method critically depends upon first constructing phylogenetic trees from genetic sequences. The dirty little secret of phylogenetics is that there are many ways to do this and they often give different answers (i.e., different trees). So it's reasonable to ask what effect the tree constructions has on the results. Here's how the authors express this simple idea:

Methodological congruence was tested for by rerunning the DELTRAN migration analysis on hemagglutinin and neuraminidase tree topologies derived via (i) a gamma-corrected maximum likelihood (ML) general time-reversible model and (ii) a distance method based on neighbor joining (NJ) and a gamma-corrected Tamura-Nei substitution model.
All this information is useful to specialists who want to know the details, but considering the authors relegated a lot of information useful to the general reader to the Supplementary tables, one wonders what the purpose of this kind of technical detail is in the main text meant for a more general audience of flu scientists.

This is a pity, since there are some interesting statements here and there. For example, the authors concede that it is well known that H5N1 came out of southern China, but that their analysis adds a finer resolution, pinpointing it more precisely to Guangdong province in the south (adjacent to Hong Kong). They then observe:

Guangdong, along with much of southern China, hosts a combination of circumstances that apparently promote H5N11 diversifiction and spread. These include explosive growth in the production of factory farm poultry and free-range ducks; extensive use of vaccines on industrial Galloanserae [geese, ducks, quails, pheasants, and relatives]; an expanding interface between wildfowl and domestic birds brought about by reduced wetlands and a newfound wildfowl penchant for human agriculture; and, with Hong Kong's reintegration into China, greater access to international trade. [internal cites omitted]
Interesting observations, but neither new nor the result of their analysis. They speculate further on general evolutionary mechanisms suggesting that the virus switches between different genetic combinations to allow it to infect various hosts species in the same and different localities. This may be true, but at best it is only consistent with but not a result of their analysis.

This may turn out to be a very good and important paper. I admit I have a hard time judging. But it could have been so much better if some reasonable effort had been made to communicate it clearly. It wouldn't have been hard nor would it necessarily have taken much extra space.

This isn't an uncommon failing in scientific writing, unfortunately. But it's still a failing. And it prevents a good paper from having the impact it should and makes this reader wonder if the technical jargon is hiding less than meets the eye.

End of rant.

my my