I was skimming through my RSS feeds in search of a "paper of the month" and I came up short. It was rather disheartening actually - a lot of current publications in bioinformatics seem to consist of:
- new algorithms without practical application
- findings of low general interest by beginners using the most basic of tools e.g. BLAST
- badly designed database frontends with no functionality
I'm beginning to worry that bioinformatics is in danger of failing to live up to its promises. We have to convince the unenlightened that our tools, applied intelligently, can provide meaningful insight into real biological problems of fundamental interest. Yet I see little evidence of this at the moment.
Could someone please find a really good paper or suggest a fantastic collaborative project, or I'll get really depressed.
Comments
Apparently you are
Apparently you are pessimistic and should be reading refereed journals (and not RSS feeds of refereed journals).
See what you have failed to appreciate is that while the current crop of bioinformatics papers are focusing on mundane things like sequence alignment methods and silly genome comparisons. In the future things will be different !
We will be able to plug undergraduate students directly into computers and have them manipulate sequence processing pipelines with the *power of the mind*.
Bizarre...
Not just pessimism but "engineer pessimism". That's really serious.
It's a strange place over there isn't it. I don't think I want to go there.
Many "strictly
Many strictly bioinformatics journals are rather uninspiring of late. I was disapointed to see the latest issue of BMC Bioinformatics includes a number of articles on low level microarray data processing. Bore yourself stupid with Sources of variation in Microarray data and Removing dye biases from microarray data, where the authors tested 41 methods of dye bias removal (pity the grad student who did all the work, the horror).
All is not lost though: Open XML architecture: a reliable and easy way to publish chemical information. Written by the guys who did Chemical Markup Language, they propose a way of embedding chemical information in scientific papers. While I'm sure you're all sick of me banging on about XML and the semantic web I still believe the goal of not spending most of your time massaging data is worthwhile. XML and the semantic web are the most likely to succeed in achieving this (I will be posting some semantic web stuff shortly, with more details).
Other picks:
MicroRNA is still "hot", and I would say that low level microarray data analysis is definitely "not" :) And why is it PLOS Computational Biology and not PLOS Bioinformatics ? Does it imply that processing biological information is somehow not an attempt to understand biology ?
As for a collaborative research project, I think that it will need some further cogitation (and the right timing). In considering this a little further I believe collaborative science is only superficially similar to open source software development. This is because the end goal of software development is software, whereas the end goal of science is new knowledge. New knowledge is arguably a lot more difficult to produce (?) than cloning a popular software product or implementing a well known algorithm. Layer on top of that issues of credit, authorship, funding, publishing and the general bitchiness of scientists and I have a feeling that some ground rules will need to be laid down first.
If I don't get to annoyed with science in the next week or so I'll try and write down some of the points that came out of the previous discussion and the work on a proposal for a suitable project... we'll see.
Thanks for saving Neil from depression
To me, working with networks for too long, many of the recent network papers are binned with the microarray normalization papers. However, there is interesting work coming with the chimp genome (this one for instance).
Concerning the PLoS journal's name - I would have chosen computational biology too. After all, (this part of) bioinformatics wants to contribute to biology rather than computer science.
Talking of networks...
Most of the stuff I seem to come across applies static network theory to biology. Have there been any dynamic attempts? I get the impression that dynamic networks are terra incognita, so noone knows what's going on. Any recommendations for a read?
Dynamic networks
I agree that too little has been done to get something useful out of these big biological networks. Most people keep saying that we should take into account things like protein concentration, affinity, dynamics, localization, but few studies have come out yet on this. I remember the party and date hubs from Vidal's lab and a similar idea from the Boork group here in the EMBL. Both studies try to integrate mRNA expression with protein-protein interaction data. A different type of dynamics, would be evolutionary dynamics. Here I would point out to the work of Andreas Wagner, that has tons of interesting (but not very visible) papers on the subject. Here is an example.
One last comment on network studies. Do you remember Albert Barabasi, the author of Linked and several papers on network properties ? He is moving to the Dana-Farber Cancer Institute, home of Marc Vidal's team , for a year. That would be the combination of one of the most well know theoretician on network studies with some of the most advanced high throughput network mapping. Maybe something interesting will come out of that :).
Thanks
Thanks for that, Pedro. I'm familiar with the people you mention (although I didn't know Barabasi was moving to DFCI). You are, as usual, right - combining ppi and expression data is *probably* a sane idea. But how stable are expression results, and can you predict phenotypes based on these changes in the transcriptome? That's where I'm going with this...