Grid Computing in Pharma - Bioinform (3/24/06)
March 30, 2006
The March 24th issue of Bioinform (subscription required) has an article on distributed computing entitled “Distributed Computing is Alive and Well on Pharma Desktops Despite Some Licensing and Tech Glitches”. Being somewhat skeptical of the usability of distributed computing in pharma environments, I was very curious to see what the article concluded.
According to the article, J&J, GSK, Novartis, and Sanofi-Aventis are all using the GridMP platform from United Devices (I was aware of three of those). Novartis, for example, is using their grid for virtual screening (I saw a presentation on their efforts a couple of years ago) with some degree of success, especially since there are no real IPC requirements in docking experiments.
In general, grids are being used for mining and similar applications, which do lend themselves well to using spare desktop cycles. I am still not completely convinced that grids are efficient enough to be used for serious crunching, especially when there is a time crunch.
The bulk of the article focuses on issues with licensing, which I whole heartedly agree with. Software company licensing models have not evolved with the times, partly due to a lack of market pressure, and partly due to an inability to identify just how to leverage modern HPC. Most licensing models still date back to the days of individual workstations and could really benefit from an upgrade.
In the end, I would agree that in certain cases distributed computing (using free desktop cycles) is a viable alternative for some tasks. If I were an IT manager, grid computing would be deployed to generate data, which can then be accessed when required for a specific project.
Technorati Tags: Scientific Computing, High Performance Computing, Grid Computing, United Devices, Software Licensing
Wordpress Widgets
March 30, 2006
I am looking forward to implementing WordPress Widgets. As in other cases, I am holding out on palying around a lot with the blog design for now, as I hold out for the next release of K2, which is widget ready.
Nature Magazine: The future of scientific computing
March 23, 2006
Nature magazine has a complete issue on the future of computing. Many of the articles are free thanks to sponsorship from Microsoft, who also have their own 2020 report. The issue makes fascinating reading, and talks about some of the major challenges that we need to address using scientific computing.
The first article I read was the article by Roger Brent and Jehoshua Bruck on computers and biology. The other article that caught my eye was the one by Declan Butler on pervasive computing. Over the next few days I hope to write some reviews on the articles. In the meantime, the articles are free. Read away.
Further Reading:
Technorati Tags: Nature Magazine, Science, Scientific Computing, Microsoft
Technorati Tags: Nature Magazine, Science, Scientific Computing, Microsoft
Education extends to computational scientists too
March 16, 2006
I have recently discovered the blog at the Open Science Project, which I heartily recommend. As I was browsing through some of the earlier posts, I found a post discussing the lack of good software practises in scientific computing. From personal experience, both as a computational scientist myself and from seeing a lot of academic code, I couldn’t agree more. When I was getting my feet wet in computing, all the coding I knew was mostly self taught, with limited awareness of good programming practises other than writing reusable functions and trying to stick to standards. I know many other computational scientists, whose knowledge of anything other than emacs or vi for writing code is rather limited, with almost no knowledge of version control. In general, the originator of the program remains the only person who can comprehend the code itself. While this is changing due to sites like SourceForge, in general scientists write sphagetti code that would drive most code reviewers batty.
What is the solution? I have argued in the past that one way for molecular modeling and informatics to be part of every lab in the world, it needs to become part of the curriculum, just like traditional lab work. A similar argument can be made here that everyone interested in becoming a computational scientist needs to take some classes in software programming and learn the basics of modern programming practises. This is a win-win situation that benefits the developer and the end-user and also greatly helps commercialization of good ideas.
Technorati Tags: Education, science, programming, computational science, scientific computing
Blind hamsters in past tense
March 14, 2006
Dr Rutledge Ellis-Behnke and co-workers have published a paper in PNAS (still waiting for the thing to show up on the PNAS website and on Pubmed), which has garnered a lot of press today. The team, which included scientists from MIT and Hong Kong University (Dr. Ellis-Behnke has a joint appointment), was able to restore the vision of hamsters using a clear solution of nanofibres. The peptidic nanoparticles were able to self-assemble and bridge the gap between the optic nerve of the hamsters (the nerve had been articially severed).
I have seen some video clips and a presentation by Dr. Ellis-Behnke of the nanoparticle solution in action, although the focus of that talk was on homeostasis. What I saw was certainly fascinating and I can’t wait to read the paper to see if it says any more about the material and its mechanism of action. The multiple capabilities of this material sound almost too good to be true, but I have seen some evidence in action, and was very impressed.
Large scale animal studies are currently underway.
Further reading:
Technorati Tags: Nanomedicine, Neuroscience, Vision, Nanotechnology, Nanobiology, Rutledge Ellis-Behnke
