PredictProtein Machine Image

From Rost Lab Open
Revision as of 15:23, 8 March 2011 by Schmidb (talk | contribs) (Example)

INTRO

The PredictProtein Virtual Machine Image (PPVMI) is a self-contained solution for protein feature prediction. The image contains everything you need to get started: a fully functional Debian system, prediction methods and supporting databases. The image may be used in small scale analysis on a single machine (real or virtual) or in large scale high-throughput analysis on an arbitrary number of server instances in the cloud. You may extend the image with packages (such as gridengine-client) and tools according to the needs of your computation environment.

Most prediction methods on the image are integrated into predictprotein(1). Please refer to predictprotein(1) for a list of the included methods.

The following methods are provided in addition:

  • snapfun(1) - a method for evaluating effects of single amino acid substitutions on protein function

Availability

This image can be downloaded from the PredictProtein website (registration required).

Booting the Virtual Machine

First make sure to unpack the downloaded image. Make sure you have at least 20Gb of free space on the device

 gunzip pp_machine_img_201103021219.vmdk.gz

KVM

Booting on kvm

 kvm -m 2048 -hda pp_machine_img.qcow2 -net nic -net user,hostname=n-1232.amazon.com,hostfwd=tcp:127.0.0.1:10022-:22 -usbdevice tablet

KVM's documentation:

 man kvm

VMware Fusion for MAC OS X

First you would need to setup a new virtual machine

  1. Create New Virtual Machine: File -> New -> Continue Without Disk
  2. Pick Use an existing virtual disk
  3. Name your virtual machine
  4. Give at least 2GB of RAM to your virtual machine
  5. Finish
  6. Boot your new virtual machine.

More information is available on the VMWare website

Virtualbox

First you would need to setup a new virtual machine

  1. Create New Virtual Machine: Machine -> New
  2. Name your virtual machine
  3. Choose Linux and Debian as operation system
  4. Locate and select the VMware .vmdk file as your boot disk
  5. Give at least 2GB of RAM to your virtual machine
  6. Finish
  7. Boot your new virtual machine.

More information is available on the Virtualbox website

Example

As an example you can run the following command:

 predictprotein --seqfile /usr/share/predictprotein/example/tquick.fasta --output-dir /tmp/$USER/pp

The results you can find in /tmp/$USER/pp. To switch of your virtual machine use the command

  sudo poweroff

Output format

All Rost Lab methods give a brief description of their output format in this section. For further details please refer to the references and the source or script of the tool.

References

Laszlo Kajan, Yachdav, G., Burkhard, R. (2011). High-throughput protein feature prediction in the cloud. Bioinformatics (submitted), XXX(X) XXX-X.

In case you find the PPVMI image and the tools within useful please cite:

  • the references for PPVMI, see above
  • the references for PredictProtein, see REFERENCES on the man page predictprotein(1)
  • the references for the tools you used, see REFERENCES on the man page of the tool

Questions

Please see the FAQ section or contact the lkajan <at> rostlab <dot> org

Copyright and License

ACADEMIC SOFTWARE LICENSE AGREEMENT FOR END-USERS AT PUBLICLY FUNDED ACADEMIC, EDUCATION OR RESEARCH INSTITUTIONS FOR THE USE OF PredictProtein

Please refer to /usr/share/doc/ppvmi/copyright for the full text of the license.

Commercial users or users to whom the above license does not apply please contact Biosof Sales <sales@bio-sof.com>.