Der Bioinformatiker

I made a panorama from my balcony in Heidelberg. It really was a beautiful day today. I went down to the city center with Elke, and we even found something for nostromo: this nice sculpture of an alien!

Finally, I found some time to update my blog. Much has happened, not all of it really worthy to write down, but I will give my audience a short update what is now going on in my life.

First, a short technical note: since my last security update, I seem to have screwed the newsfeeds. I hope I will find some time soon to fix them.

My work here at the EMBL Heidelberg is nearing its end. Today, I managed to get the major part of my program working, now I’m much more relaxed. All I have to do now is some fine-tuning and data collection for my thesis.

I guess I’ll explain what I am working on here. It is not classical bioinformatics, at least not how I solved the problem The group I’m working in is researching in the field of Cryo-Electron-Tomography.

• Tomography is a technique to get 3D data. It is based on taking transmission images (with X-ray for humans, for example) from different angles (preferrably evenly spaced around the 360°). This data can be reconstructed to a 3D density model.
• The “Electron” signifies the use of an electron microscope
• “Cryo” is added to point out the sample preparation technique: the sample is shock-frozen in water, in a way to get amorphous ice, which is frozen water that did not form ice crytals and therefore does not destroy the sample

The resulting technique allows us to generate 3D images of a cells interiour, in a next-to-natural state and high resolution. The resolution is high enough to detect bigger proteins and get good data on the exact organization of a cell.

My work is/was to work on the reconstruction. I get the projection data and feed it to my program, which then calculates (and calculates and calculates…) to reconstruct the 3D structure. Because it is so computationally expensive, parallelizing the work is a solution to get the results faster. Although we have access to a big cluster of high-performance machines, it would be nice to use these for other parts (for example the things Bernhard Knapp, the other bioinformatician from Hagenberg here, is working on – finding and clustering the particles in the data) and fortunately we had a great idea – why not use the graphics card? (Well, to be honest, the idea was found by another company, but their product does not quite fit our needs…) – so I researched a lot and started programming. My first prototype (which we tested before easter) was around 10 times faster than the calculation on a CPU. Now, I finished the second prototype and I’m about to run some speed tests… I hope that it will be not much slower now. If anyone is interested how this speedup can be reached, take a look at the GPGPU website or ask me…

Mein Balkon

Originally uploaded by Hannes Müller.

Yesterday was a great day, all sunny and warm. So I sat in my dark room and chatted in IRC, complaining about how much I would have liked to sit out in the sun. So JG^ told me “IKEA is still open”…
One hour and 15 minutes later, I’m back with my new table and two chairs. I even got a candle for those extra romantic nights…
The only thing I got to figure out is how to read the laptop screen in the sun.

Well, enough of these trivialities. I guess the next blog entry will be all techie and stuff again…

Skiwetter
Originally uploaded by Hannes Müller.

Elke and I went skiing on April Fools Day, for the last time this year… The weather was great, only the snow was a bit heavy because of the warm temperatures. I put some photos on Flickr, just klick ahead…