Skip to main content
Topic: About multi-cores use (Read 1050 times) previous topic - next topic

About multi-cores use

Hi,

Im wondering how XCMS can use all the cores available on my computer.
During an XCMS processing, only the peak picking step seems to be able to use several cores (thanks to nSlaves). In my case, the 'slow' step is the retention time correction (obiwarp method) and it seems that it use only one core to do the calculation. Assuming that obiwarp compares the center sample to sample 1, then to sample2...etc i'm wondering if it's possible to use all the cores of my CPU?

If it's already implemented (sorry for this post!) what is (are) the parameter(s) to add to the command line?

Thank you for any precision,

Best regards,

Musca

Re: About multi-cores use

Reply #1
Only the feature detection step supports parallel processing at the moment.
Set the nSlaves parameter to the number of cores in your workstation, e.g. xcmsSet(..., nSlaves=4)

What step size do you use for OBIWarp?
OBIWarp's speed mainly depends on the step size. profStep=1 should be reasonably fast and gives great results.

Ralf