Notes from the Biomass

... will be the headline when this essay from the editorial of PLoS Medicine winds up in the mainstream-mainstream media. It is already entitled Why Most Published Research Findings Are False and quoted as Most scientific papers are probably wrong by the New Scientist.

The work from John Ioannaidis, an epidemiologist at the university of Ioannina in Greece will stir some controversy and will be water on the mills of the quacks. The paper appears solid but most readers with some statistics experience won't be surprised, let alone experimentalists dealing with interpretation of results of complex experiments.

Does this editorial really say something that is novel and that has to pronounced as dramatic? Not that we should bar any scientific work that is peer-reviewed (are the editorials peer-reviewed?) from publication but does it really need a title like that?

On the other hand, a lot of research is sold as if the cure for cancer is just around a corner and a dose of scientific reality certainly won't hurt.

[Thanks, gruggled]

Two phrases dominate the label scientist use in bioinformatics. Couldn't the definitions people use for they "job descriptions" be labeled escapism?

• Bioinformatician - A scientist with a background in biology or biochemistry who stopped generating data himself
• Computational biologist - A computer scientist or physicist who happens to use biological data

Seriously - go to any conference, ask the participants for both their "label" and their background. I got 100% correct (sample size:2). Am I a biostatistician now?

Jason over at The Personal Genome found an ... ahem ... interesting prediction on the future of medicine by Leroy Hood in the Seattle Post. Instead of depleting my stock of Niels Bohr quotes, I rather point you to Jason's comments.

Preparing for coming bioinformatics conferences, I shudder to think of the talks that sound great in the program but turn out just to be the nth naive bioinformatician. Five things I really don't want to hear again this (but probably will) later this year.

1. Reductionism and Better understanding used in conjuction with biologist, who supposedly don't get the whole picture because they only study one protein in their life. Used by computer science graduates that just downloaded two different data sets from the web, whacked it through R and think the biologists were just waiting for it.
2. Overviews that consist of introduction, methods, results, discussion and conclusion but no word on the subject.
3. Pointing out groups of genes as interesting and ask for experimental validation thereof. Every gene is interesting be it conserved or specific, abundant or crucially regulated. The most boring gene would be highly interesting, wouldn't it?
4. Complaining about errors in public data bases. A common theme in presentation from people who were incapable of restocking the pipet tip racks are pointing out errors in annotations or relabeling of genes. Such behaviour should be punished by taking those peoples code and highlight the errors in their comments (or lack thereof) on the conference web site.
5. Names of proteins - If you've been working on isolating a protein it's your undisputed right to christen it. If you want to call it after the teenage mutant ninja turtles or simply p4.1, go ahead. If you have named your variables $aaa, MainClass.java or britney.h you are just as guilty anyway.

Thanks, but no, thanks, Dr. Freeman, I have no use for your crowbar now that I wrote this rant.

Some days back, the archive of Proceedings of the National Academy of Sciences (of the USA) was extended back to 1915. Browsing the back issues tells much about how science was done back then. Publication style changed quite a bit over the years. The canonical methods/results/discussion separation was not enforced back in the 20s and the publications read more like letters that the highly standardized, specialized publication we get today.
Also, the short titles of the single author papers show how it was possible to be achieve much without running being part of larger group.

It would be nice if all journals would open their archives this way. Nature is (only) going back to 1950 and requires special licenses for issues from 1996 back. Science has not made his back issues available before 1997 - if you want to read about the genome of Haemophilus influenzae, you'll have to order it.

For the bioinformatic text miners, the character recognition and access to the whole corpus might remain an issue. It will be nice to have the whole corpus of scientific publications under one roof - it won't take much more than a couple of years to complete.

The genome of the peculiar oceanic bacterium Pelagibacter ubique has been sequenced and is presented in the current issue of Science. The bug contributes massivly to the biomass in the oceans, owing in part to his effective genome markup.
No phages or transposons, few paralogs and no recent duplications or pseudogenes were found but the ability to synthesize most metabolites, including the 20 aminoacids. This bug must be treasure trove if you are studying metabolic networks. All is coded neatly into 1354 genes separated by a median spacer of only 3 nucleotides (!), making P. ubique the smallest free-living microorganism.
Somehow, I like this fellow.