Comments on: Akademy: my own BoF

By: Stu

Stu — Thu, 03 Jun 2010 16:00:42 +0000

Looking forward to discussing KDE and science* at Akademy

* in the general sense – what I know about bioinformatics is probably too small to be detected by the most advanced scientific techniques :-)

By: Einar

Einar — Sun, 30 May 2010 18:25:44 +0000

Plotting and visualization widgets would be best, as it’s essential when you display results to a biologist (who is the one that will ultimately decide if the results are meaningful or not). For my own field, good widgets would be ones to display box-plots or heat-maps, as they’re widely used.
For chemistry and the like, Avogadro is already a great start (and it’s Qt-based, even!).

By: armijn

armijn — Sun, 30 May 2010 17:52:34 +0000

(gosh, it appears I have a follower. I’m honoured)

I do agree that mainframes are not the right solution for everything. I know of a university here in .nl (Wageningen UR) where there are several clusters running. All of these have been tuned to fix a different problem: some of the programs are CPU bound, others are memory bound, others are disk bound, so they all need a different solution or else you will be wasting time.

But, that’s a bit of a side issue. Another thing I was thinking about is that it would be good to investigate what is currently a big problem in bioinformatics and how we could remedy that. For example, what is the biggest user interface element or class they could benefit from? I wouldn’t be surprised if it would be some component which can visualize HDF data, or which could visualize things like molecules in an easy way. With something like that you have a real selling point.

By: Einar

Einar — Sun, 30 May 2010 16:19:27 +0000

@Edwin: The vast majority of web based applications are non-Free, and this brings huge problems on their maintainability, especially when due to lack of funding the services shut down (I had this happen to me twice).
A FOSS, desktop based application is certianly better because it does not force you to depend on someone else’s systems, and does not force you to have a network connection. There are good examples of applications like this, such as Cytoscape.
Also, web based front ends don’t work that well if host != localhost when you deal with data on the size of hundreds of microarray files or high-throughput proteomics experiments. I would certainly not use HTTP for experiments of the size I’m working on at the moment, for example.

By: Edwin

Edwin — Sun, 30 May 2010 16:10:00 +0000

I run a Proteomics facility and use a fair amout of bioinformatic tools. They tend to come in 2 forms. Windows based apps which are commercial and often very buggy and more focused for small facilities, and open source tools more suited which are generally command line driven but have the huge advantage of being able to be supported in a computing cluster.

I’m not sure what role KDE could play in all this. Most work for developing front ends to make the use of opensource tools easier is targeting web based apps, so KDE would not play a role here.

By: Einar

Einar — Sun, 30 May 2010 09:38:38 +0000

This can’t really work with small research groups, which do not have the resources for such an infrastructure. Plus, you don’t need a huge computing power for everything. Lastly, the use of a platform like KDE may help making some of the routinary tasks more easily available.

By: armijn_follower

armijn_follower — Sun, 30 May 2010 00:25:08 +0000

I do not think that KDE will assist bioinformatics research. The need to use assembler language (second generation computer language) can be a daunting task for non-programmers. That same need would be diverting the sole purpose of research in bioinformatics. I do not know any computing machine that can write algorithms, but for data, I suggest mainframe computers because they surpass the volume of personal computing machines.

By: Einar

Einar — Sat, 29 May 2010 23:56:56 +0000

Well, I did not mean that pros only use C++. ;) The real explanation is a bit more complex, but the entry was already getting too long…

As your other questions, they are very valid points. The answer to some of these lies in the (very complex!) interaction between the bioinformatician and the biologist, a love-hate relationship.

Also, the use of an established framework helps to solve a lot of problems related to software in research. You don’t get funds for maintaining software, just to create it, and established frameworks limit the effort. Secondly, there’s a massive NIH syndrome in bioinformatics, unless you’re truly a trend-setter. KDE offers a lot already as a development platform, and would reduce such problems (also excellent applications like rkward or Cantor show how good can KDE be for science…).

Anyway, you gave me good food for thought to explore this even further. Thanks.

By: armijn

armijn — Sat, 29 May 2010 22:14:05 +0000

Funny. I never knew that to become a professional coder you need to know C++. I think you will find that a lot of “professional” coding is done in other languages than C or C++. So, don’t feel ashamed when you code in another language :-)

Apart from that, I think that it would really help if the BoF could focus on answering a few questions: what problem will KDE help solve for bioinformatics researchers? Why would it be better than other solutions that are already used? How much effort would it be to switch? Without answers to these questions it will be very hard to get people to switch.