Welcome, Guest. Please Login or Register
UGENE Bulletin Board
  Welcome to our forum.
Page Index Toggle Pages: 1
remote execution of commands (Read 5510 times)
Apr 29th, 2011 at 10:01am

Poh   Offline
YaBB Newbies

Posts: 46

I got a simple read and write schema written. It works locally, read a file call test test test to write a file test test test.

I am trying to run it remotely. I am not sure I got the remote configuration right. is there a way that I can ask for remote service somehow?

I will be testing another one with blast+.



ugeneremote.JPG (181 KB | )
IP Logged
Reply #1 - Apr 29th, 2011 at 4:14pm

Mikhail Fursov   Offline
YaBB Administrator

Gender: male
Posts: 162
Running computational tasks (workflows) remotely is an experimental feature today.

We tried different methods and about 1 year ago to run a task on remote computer all you need is to run another UGENE instance. We closed development in this direction because server instance must have special logic/interface/security logic client-UGENE does not have

Today to run UGENE remotely you need a special server UGENE application that includes adapters to HPC environment it is installed in.
We have different adapters now: for EC2, for local University cluster (Novosibirsk) and can write new efficient ones if will know details about the environment.

A free experimental service based on EC2 environment is available at https://ugene-service.com and is free to use for everyone.

The protocol we use in UGENE to communicate with the server is in the source code. The server user management and task scheduling code is written in Java and is not opened now, but I hope it will be opened soon.

Any interest to this part from some organization (academic institution) to install UGENE server locally to their own environment will definitely raise the priority of this feature for UGENE team. Plus, our team will provide a help in deploying the server and optimizing HPC adapters.


UGENE team
IP Logged
Reply #2 - May 2nd, 2011 at 9:00am

Poh   Offline
YaBB Newbies

Posts: 46
Well... for large scale data we cannot afford to run it locally. Some analysis can be depressingly slow, but can leverage on HPC environment such of that provided by SGE or rockcluster (http://www.rocksclusters.org/wordpress/) that could run the job on a few nodes at a time, doubling and tripling the speed of execution.

If I could put ugene in the headnode, anyone could call it up and execute the workflow there and the SGE delegate the jobs to the worker nodes. But I am not sure if that is considered local or remote execution.

I would agree ugene may need some form of server function (ugene server?) for remote execution of task from ugene clients. But targeting SGE remote job execution if possible could be a good development direction.
IP Logged
Page Index Toggle Pages: 1