Company Visit I: Agilent Technologies
January 11, 2008.
At about 10.45am we (me and some uni karlsruhe students) departed from Infobau (the main building of Computer Science Faculty of Uni Karlsruhe). About 45 minutes later we have arrived at Agilent Technologies in Waldbronn. We came to a Conference Room and listened to some introduction from Agilent. Theres some story about the history of Agilent as HP, and then the buildings (about 5 buildings there, and the Agilent representative had said that they still have plans to build more). Agilent is an international company that have a branch mainly in USA (East Coast & West Coast), Germany, India, Cina, Japan, and maybe more countries that i can't remember. Their business products ranging from scientific instruments, semiconductors, optical networking devices, and electronic test equipment for telecom and wireless R&D and production. They researching in many fields from chemistry, biology, physics, farmacy, computer science, and many more. Well, enough for the crap, what do the researcher actually do in this branch of Agilent in Waldbronn?.Agilent in Waldbronn developing tools in chromatography. What the heck is that?. Thats a laboratory techniques for the separation of mixtures. A little example: We have an pen with ink in it, a filter paper (a long rectangle paper), an a box of water. We touch the filter paper with the pen, an absolutely a stain appeared on the paper. We take the paper, and hold it above the water box, so that it touches the water a little bit. The water will absorbed, and the stain, strangely, separated and forms 4 stains with various different colors. What happened?. The stain composed of a mixture. And the water absorbed will push it to top. The mixture composed of different molecules with different sizes. The small size molecules runs faster, the larger ones slower. Thats why we could see more than 1 stain, where each stain have a different molecules in it. Well, now, in modern time, we don't use such technique with paper. Agilent developed all the tools and machinery that make it possible to do the chromatography accurately & efficiently. In the past, a whole laboratory was needed to do all the steps in chromatography. Well, theres a new solution, a very ingenious one indeed, with the slogan "Lab on a chip". The machinery itself called "Bioanalyzer". Accomplishing the job of a lab in one single chip. Well, after a lot of presentation slides, and photos of machines, we go to one of the great reason that we come here, dinner time!!!, and its all free. I think i don't have to explain about all the delicious foods that we have there. 1.15pm, we are divided to 2 smaller groups, and the tour begins.
The first part of the tour shows the phases in actual manufacturing of the machines. The manufacturing room is really clean, modern, and quite noisy too. Theres a lot of voices of machine testing, screw driver, etc. We were shown the actual workstation & stacks of machines. Theres one quite interesting knowledge there. It had been researched that actually the most expensive part of the machine is the screws. Thats why Agilent develop the machine compactly with very minimal screws. It had the advantage of being cheaper, easier to disassemble to replace/ update the parts, and also designed specially so that just one fan needed, to cool the whole box. And then we were explained exactly how the little parts works (honestly, i still don't have a clue about it). The machine assembling room & the final assembling room (final product with the casing closed) located strategically so that no time wasted to communicate or transport things. The other rooms located with the similar idea. The product's defects must be detected as early as possible (better while not yet assembled), but theres also the final quality control. There is a special machine to test the products, and 95% of them pass the test & directly packed to the customer. The rest 5% will be fixed in a reparation room.
The next part of the tour is led by a biologist (he is a sharp young scientist that have a lot of experience includes with Fraunhofer Institute etc). The biologist explains and shows how the Bioanalyzer works. Just fill the chip with the substance to be "chromatographed". Choose the right chip of course, he used as an example the chip for RNA. And then put it in the machine (its actually quite a small machine, looks like a common size laser printer to me). And then runs the program to connect the computer to the machine through the appropriate COM. Well he don't actually runs it with the actual substance because it would be about 25 minutes until it finished, and we don't have such time. He uses demo version instead. And we could see the amazing menus (well designed user interface) in the software, its still runs in WinXP btw. The results displayed as a signal (line graph): x-axis = time, smaller time means smaller-size fast molecules, y-axis = the quantity of molecules detected, written in percent relative to the quantity of molecules of the whole substance. There are peaks in this graph that shows the molecules contained in the substance. Theres also a profile that shows how the results should be. And we could compare it, so that we know how good our substance ist (for example sometimes RNA being in room temperature too long could be broken, and so its signal will have a large amount of noise, we can see it by the peaks, besides the peaks theres should be a relativily flat signal (almost 0)). He had also told a story about a Diplomarbeit (compatible to master thesis) from one student that give ratings to the resulting graph. 10 being the best with minimal noise, 0 being the worst. And in judging between 10 and 0 he had succeed to designed an algorithm that mimicks the knowledge of the experts in that area.Next, we have enlisted our names to the project's list (its now 1 group max. 5-6 persons only). And the first one for me is "genetik algorithm in controlling the result quality". The presenter is a student too. He studying in FH Konstanz, and he have Diplomarbeit in Agilent about it. Not yet finished of course, he had just 2 months working there. So, at first he showed us the criteria of the expected result (result of which is peaks and which is not) : theres a false negative (alpha-error, most important to avoid), false positive, and the right results. The false results should be minimized with this genetic algorithm. How?. Theres a 2D array with blue and red colored little boxes as its element. Each column is one individual, each row is the parameter (thresholds, etc). Fitness function = minimize the false pos./neg. Begin with the iteration, on each iteration 2 best individuals get a free-pass to the next iteration, the rest must be married and 2 offspring resulted with some mutation, crossover, etc (common genetic algorithm rules). After about 250 iteration, hopefully, the best combination of parameter found, and so only the real peaks identified as peaks, and all the real peaks be found. The program will be enhanced with Simulated Anealing, Hill Climbing etc. Its not yet finished, so we still can't see the final program.
The next one is "Software Lifecycle". The presenter have this funny german dialect with a lot of "-isch". He had worked 9 years in Agilent. And he explains about the steps of planning, designing, implementing software. And then still have to take care of the maintenance, customer complains about bugs, etc. Theres an explanation about the "waterfall" scheme of the software engineering, a lot of talks about revision system of the software, and others that i can't remember or understand.
And at last the closing presentation. Its about Agilent on the 2nd phase (inventions, great employees, connection with universities etc), with the 1st phase of separating from Hewlett-Packard is already completed. There were a lot of talks about positions that could be filled by us in the future (programming related). And the opportunity of master thesis, parttime job, or something like that in Agilent. The most interesting, is about the talks about some seminar in the last Bioinformatics conference in England. The Airbus A380 is designed by computer to make sure its very good quality, efficiency, and other specifications. Why not drugs?. Thats one of recent holy grail in bioinformatics. To automatize the designing of drugs, and therefore maximize its quality in combating diseases.
Conclusion: A very interesting experience, can't wait till the next Company Visit!.


