Animesh – I don’t do large scale analysis anymore, so my computing needs are pretty small. But of course I wouldn’t run the whole genome analysis on my desktop computer anyway. Virtualization, from the end user point of view, removes some overhead of running software at all. If speed is an issue, maybe virtualization similar to AWS/S3 services would become more popular?

Asif, thanks for the information about these Bio-LiveCDs. As I said, both solution have their advantages, and they complement each other. Can you comment on scalability of LiveCD solution? Does it require remastering of the CD?

With regards to the “Bio IT” specialist, the server versions of our liveCD systems enable one to set up a fully operational bioinformatics node within half a day on a system with large capacity hard disks, such as a system with two 500Gbyte hard disks, which are relatively cheap nowdays, ~USD1,000). We recently set up such a server in Vietnam at the Institute of Biotechnology, Hanoi, as part of our workshop funded by UNESCO-IUBMB-FAOBMB (http://incob.apbionet.org/incob07/hanoi.shtml). With these portable and scalable systems, it is now easy to set-up bioinformatics labs, fully equipped with LAMP, Mediawiki and more than 200 freely available bioinformatics software. New software packages can always be made available by remastering the liveCD systems.

LiveCDs (or perhaps LiveThumbdrives in the future) may be the way to go forward

I recently spent some time reading up on recent developments in distributed scientific computing – not quite the same issue as “getting everything to run everywhere”, but more like “being able to run lots of things faster more easily”.

In that respect, there are a number of packages that seem similar to Taverna, like Knime and running MapReduce-transformed algorithms on Amazon Web Services. The latter is basically a way of accessing a cloud cheaply without buying any hardware. Maybe there should be a community-driven project implementing Emboss in Knime on AWS…

Sequence Alignment
BLAST
Phylogenetic Tree Building
Primers Design
Motifs and ORFs
Restriction analysis
Sequence editing
interactively view 3D molecular structures
restriction analysis
etc.

Concerning BioIT – I did think about similar approach, but I couldn’t name it so well, as you did. Hiring technical personel is already a fact in experimental biology, so maybe it’s time to do the same in biology in silico? So far I didn’t hear about such case…

I’d also like to see the day when we all use workflows and clouds, but I suspect it’s still a long way off. Let’s face it, convincing many biologists to learn how to use a new computing resource of any kind is difficult. I’ve always felt that the most effective solution is for every academic department to employ a “Bio IT” specialist – someone who provides IT support but aimed specifically at bioinformatics. Give them a few cheap servers, let them install whatever local resources people need and make sure that people know what they offer and where to find them. Need a custom BioPerl script, a local BLAST database, a Pise interface to a software package? Go to the Bio IT person. Might even convince a few researchers to extend their computing skills. Any kind of “one size fits all” solution to Bio IT needs is sure to be deficient in some way to some person with some problem, IMHO.