Do People Use Grid Computing For Ngs Analysis Or Just Local Cluster Computing (Sge/Pbs)

Question

[William]

Do people use Grid computing for NGS analysis or does everybody just local cluster computing (SGE/PBS) ?

We have acces to both but always just use the local cluster computing.

3 reasons for this:

• easy job submission

• easy data staging. Just read write to and from local linux NFS file servers.

• we don't really have a compute problem, just a lot of data :)

Are there other NGS analysis where it makes more sense to use grid infrastructure?

Analysis where you have a bigger computing problem on less data. For those analysis it might pay of to invest in using the complexer middleware and data staging so you can use more compute resources?

[Manu Prestat]

Interesting survey/question. Could you clarify why you associate PBS to local cluster? Thanks.

[always_learning]

But I understand that you can also use Grid Computing with local cluster computing. SGE is after all Sun Grid Engine only. We are using SGE on Amazon AWS for all of our NGS related stuffs.but lately we are also looking out (trying) for putting files at K luster for some cost cutting reasons.

Answer 1

[Spitshine]

From what I have seen in recent years, big machinery doesn't seem to cut in the data intensive "stupid" processing jobs posed by NGS and relatively small individual machines (often without NFS) appear to be better suited for these jobs. They can be easily replaced after a couple years.