"-p" option in cufflinks, topHat, etc.

[amcloon]

Hello,

I'm trying to figure out topHat, cufflinks, etc. and I am getting them to work at the moment (fingers crossed) but I'm trying to understand what I am actually doing. what does the "-p" option mean? In the manual it says "use this many threads to align reads. The default is 1." What are these threads and what is the advantage to using 1 or more than one?

The example workflow in Trapnell et al, 2012 (Nature Protocols, vol 7 pg 562) uses " -p 8" in each example entry, but there is no explanation for changing this from the default.

This might be something obvious to the non-novice unix user, but it is not clear to this novice unix user.

Thanks,
Anna

[fpr]

Hi.
The number of threads indicates the level of parallelism that you want to use with the software (e.g. how many simultaneous task will the program run). This number is limited by the number of CPU cores in your system (now a days, most computers have multiple cores). If your machine have 8 cores and you want to dedicate all of them for this task then set -p 8. Bottom line is that you need to find out how many cores are available to you and then decide how many do you want to use.

BTW Most multi-threaded programs have an option to specify how many cores to use. However it is not always -p, that is a convention used by the Tuxedo Suite.



fj

[amcloon]

Ah, thank you FJ, that makes sense. So, if I interpret that right, the higher the number of threads, the faster it will go, but the more it will monopolize computer resources.

[fpr]

Ah, thank you FJ, that makes sense. So, if I interpret that right, the higher the number of threads, the faster it will go, but the more it will monopolize computer resources.

Yes it would go faster if your computer have the available resources. Think of of a core as an independent worker. If your computer commit more cores than what is available you will over-subscribe the resources causing your program to slow down. This is the simple way of looking at this. There are other issues like the amount of RAM memory available (e.g. how much space a worker can take), etc. If the cores are there and you are not planning on doing something else is preferable to use them all if possible.

From your mention of Unix, I presume that you are running some kind of time shared environment. If that so, you can find your systems specifications which include the number of cores available in each computer etc.

I hope this helps. Good luck!