Notes from the Biomass

Coming back from a retreat in Oberbayern where we discussed the validity of bioinformatics as an independent science and its apparent lack of reproducability I stumble across this article. It is a little dated but certainly worth the read.

[Via Spreeblick/ Hudsonblick]

The current issue of Cell has a number of interesting comments on Systems Biology.
Edison T. Liu's sounds a little funny to me.
<quote>"The greatest challenges in establishing this systems approach are not biological but computational and organizational."</quote>
As long as the high-throughput data is as messy as it is, assembling the systems will be a herculean step. The computational problem is a mere laugh, if you ask me.

There is a Northern Europe series of Bioinformatics conferences - this year it's in Estonia and the program sounds better than ISMB or RECOMB.
I only learned of it recently - haven't decided whether I go (getting there seems to be some kind of a drag) but it's good to see other conferences emerging.

Well, not that this would be big big news - but I played with the 2.1 version of Cytoscape, a tool for network layout and analysis and I am impressed about the changes.

I thought the old interface was hard to use and not very userfriendly - but the new release from February 2005 def. has changed things to the better and did not undergo feature bloat.

Check it out - particularly if you were disappointed with the 2.0 release.

It might be interesting to compare the data from back then - for the odd publication in recent science history.

More interesting is that Slashdot's post ranks highest in Google news and seems to be ahead of my other sources. To me this seems more interesting than the data, as there will not be much to be gained - that data is a couple of years old and the finishing work on the individual genomes as well as the advances of Ensembl will probably means no improvement to the data quality for practical purposes.

The best Celera gets out of it is a better page rank and the academic community gets the satisfaction of a failed business model.

But you know: Industry won after all: The DNA sequencers in the public genome project came from Applera, Celeras mother company and they had to invest quite a bit in order to finish with Venter.

One of the puzzling, recurring utterances of bioinformaticians young and old are the complaints about experimental data sets. The remarks that are errors in most of the biological data we are dealing with are abound.
I recall the former physicist who complained about the change of a few genes in yeast and the Postdoc who complained about the mRNA expression data who made cheap jokes about his collaborator - and any conference data has at least three tables that discuss such issues too loudly.

Actually, they should rejoice that there are errors - if the data gets fairly clean, the failure of many bioinformatic predictions will again shift into focus. The precision in structural information on proteins has not solved the fold prediction problem and the nice and precise information on eukaryotic genomes has not made de novo gene finding a routine task.