Low budget workstation for analysing bacterial Illumina/454 sequencing data

[shenge]

Hi,

I would like to build a workstation that is capable of analysing Illumina and 454 sequencing data from bacterial samples on a £2000 budget. I am looking into parallel processing using GPUs, but any advice on which CPU would be suitable or how much RAM and hard disk space would be required would be greatly appreciated.

Thanks in advance.

[GenoMax]

Hi,

I would like to build a workstation that is capable of analysing Illumina and 454 sequencing data from bacterial samples on a £2000 budget. I am looking into parallel processing using GPUs, but any advice on which CPU would be suitable or how much RAM and hard disk space would be required would be greatly appreciated.

Thanks in advance.

There are several past threads covering this topic on SeqAnswers (http://seqanswers.com/forums/showthread.php?t=30988). The CPU/GPU topic has also been covered (http://seqanswers.com/forums/showthread.php?t=33622).

Other than RAM (48-64 GB, if you are primarily going to work with bacterial data) it would be hard to get a consensus from recommendations you will get from others.

You probably won't go wrong with most any hardware you buy that fits your budget (only give or take will be the time needed to finish the analyses depending on the hardware you get).

[shenge]

Thanks for your help. I am pretty new to this and I'm currently working with metagenomic data, taking the raw sequencing output from 454 sequencing then OTU picking, assigning taxonomy and conducting diversity analyses using the 16S gene. However my supervisor would like to start comparing more genes via Ilumina sequencing and our current workstation wouldn't be able to handle that. Would using a GPU be beneficial for this kind of analysis or would I be better getting the best CPU and RAM combination I can afford?

[Richard Finney]

Here's one take on the tradeoffs (from Intel's Throughput Computing Lab) :

http://pcl.intel-research.net/publications/isca319-lee.pdf

"In the past few years there have been many studies claiming GPUs deliver substantial speedups (be-tween 10X and 1000X) over multi-core CPUs .. we perform a rigorous performance analysis and find that after applying optimizations appropriate for both CPUs and GPUs the performance gap between an Nvidia GTX280 processor and the Intel Core i7 960 processor narrows to only 2.5x on average."

Note the GTX is a fairly hefty GPU ($650 at newegg ?).

Since little code is actually written for GPUs, the CPU only solution is probably the best solution. You're best off using as high powered (multicore) CPU (i7, xeon) as feasible and using large, fast RAM. The big determinant of speed improvements is no longer higher clock frequency [ which has flatlined last several years] , but RAM/cache speed.

GPUs remain an intriguing field of research and very popular in the video games realm. Regardless, unless GPU algorithm development is your specialty, it is best to stick with the baseline, general purpose configurations.