Notes from the Biomass

Greg Wilson, author of Data Crunching is offering a course on coding practices in scientific software development. There is also a quick survey to determine whether the course is for you. However, I guess this is recommended reading for the many bioinformatics scientists who are struggling with day to day problems as programming is only one of their activities.
The course is written for people with basic understanding of Python.

With all the changes of the way people using the internet for commerce, such as they are discussed by Steve Rubel here - what happens to the sciences and the way we communicate it?
Initial, scientists such as Newton and Leibniz exchanged letters directly; later, periodic journals published those letters. These publications turned out to be a profitable business and finally the publications and their citations have become the prime the source for the evaluation of scientists.
Recently, initiatives such as the Public Library of Science, backed by established scientists and funding changed the way we think how these publications should be payed for.
What's missing to me is the exchange of ideas and communication using the internet. Of course, there are chat rooms, most scientific programs from academia are released under one open source licence or the other and available quickly (not that they are usually maintained but that's a different story).

However, the way science is performed has not changed much: Work is usually not communicated to the outside, unless publishable in a journal with appropiate impact factor. Yes, there is email communication but that's only within a closed circle.
Good ideas and important results are not placed on a scientists website but get under a blanket until the PR and the law department of the research departement agreed to - and publishing on ones web site usually prohibits publication.

In mathematics and physics, preprint server have taken a role that might be closer to what I think how scientific communication would occur if we would not have the publishing legacy and would communicate more freely. Biological sciences on the other side have not ever come close to such an open way of publishing.

The internet has changed scientific discovery completely - but the way we communicate the results feels ancient. Even the little things, such as the one-dimensional author list, which does not recognize inidividual contributions in any way, appears to be carved in stone. As science is funded top down and there is little possibility to be successful with a handful of smart ideas and fund your own research, will there be change? Or is the change already there and I am just a little annoyed about some decorating relict?

Connotea, a service provided by Nature lets you store your references and share them with others for collaborations. Note to self: Should give it a try in the next try external project.

BoingBoing just announced that papers generated via SCIgen were in fact presented at the WMSCI 2005 conference. Kind of. I found the paper generator pretty original to begin with and it has been in the list of noteworthy links for some time.

Research groups from Texas A&M and the University of Illinois studied the chromosomal rearrangements in the evolution of mammals. Their study appeared in Science.

Rates of chromosome evolution within mammalian orders were found to increase since the Cretaceous-Tertiary boundary. Nearly 20% of chromosome breakpoint regions were reused during mammalian evolution; these reuse sites are also enriched for centromeres.

Nice paper ... and 124 pages supplementary material.

The current issue of Bioinformatics starts of with the very useful review of the evaluation of clustering results.

Clustering is easy: You always get a result that can be called a success by one measure or the other. The authors of the review consider mathematical properties of the solutions and show how to compare them amongst each other.

It is a handy reference for your next clustering problem - with all the high (and medium) throughput data pouring out of the labs into the bioinformatics offices, there is a steady interest in such techniques. Even if clustering often fails to explain much of the data.

The Chronicle of Higher Education features an article, titled Bloggers Need Not Apply, which is an interesting - and scary - read for the academic job seekers. I wonder whether the board, who was disappointed by the content of the blogs for one or the other reason would actually think that the applicants that did not blog were more capable and did not possess such flaws. Better the devil you know...

I've read a number of silly rants where young scientists commented on their PIs - doing that is stupid to begin with. Most science blogs that I read do not talk about much about the daily grind but reflect on the field in general, which I miss with many scientists that focus on the few angular seconds of their day sky.

A well kept blog will add to ones professional standing - certainly, there are a number of pitfalls one has to watch out for. Well, if I didn't, I should stop immediatly. Anyway, I might want to print the article and put it under my pillow.

[Via malorama]

The notion that many networks were scale free was an important discovery and relevant to all the people that studied protein interaction networks, obviously. Many people in my circle of peers, including me never liked the idea and always found it overrated.
Berend Snel and Martijn Huynen, then at the EMBL, noted that the scale does not really span the many orders of magnitude required to actually make the statement that the distribution is in fact independent of it. The most connected protein, JSN1 in yeast had 344 interactions (within a single data set), which can't compare to the millions of links one would find in the topology of the e.g. the internet.

Nevertheless, many bioinformaticians put the notion forward, and many articles included "scale free network" just because it sounded sexy.

Recently, a group around Marc Vidal published a simple, yet revealing experiment, to identify, whether we could acutally observe such properties given our incomplete and patchy sampling of protein-protein interaction networks.
They simply generate random networks, sample from them and were able to identify scale free properties in all of them.

The networks might be in fact scale free but I do not think that the notion will tell us much about biological networks, in particular as we cannot distiguish between sticky proteins that simply bind to a large number of proteins without transmitting any signal and methodological artifacts, both largely contributing to the proteins with many connections.

Pubcrawler used to be my alerting service; I followed important fields, collaborators, competitors using canned queries at Pubmed and receiving results via email.

After I started blogging, I used RSS readers; it did not become an instant love-on first sight as many RSS readers struck me as bloated and annyoing. However, the Sage extension for Firefox works well for me - no silly icons jumping in my Dock or Panel.
Still, I never got the hang to use HubMed, but after the recent announcement of Pubmed to offer RSS feeds, I decided to give RSS feeds another try and whacked a couple of my important queries into Hubmed and Pubmed and compared the results to the Pubcrawler queries.

While it's a little early for comparative results, but doesn't appear as if the RSS feeds offer much benefits over the good old Pubcrawler - you can't track old results easily, Hubmed often highlights old information as new and Pubmed does not give you previous results, only the newest ones.

Seems like I won't change any time soon - and Pubcrawler has that personal touch to it that makes it fun to use. I should get to an Irish pub soon and drink to it.