Why is my ADMIXTURE random seed always 43
1
1
Entering edit mode
5.2 years ago
devenvyas ▴ 680

Hello, I have been running replicate analyses with ADMIXTURE. They always come out with the same CV value and similar likelihood values, so I was wondering what is the problem.

I have since come to notice my random seed is always 43. Isn't the random seed supposed to change by default. Does anyone know how I can remedy this? Thanks

admixture -j5 --cv filename.bed \$K
2
Entering edit mode
5.2 years ago

I think you'e misunderstanding what a random seed does. A random seed can given to a program to ensure that the pseudorandom number generator always produces the same series of numbers. This is useful for reproducibility and testing. If you don't want this, programs generally allow you to set no seed (which means that it's initialized from some other source, like the milliseconds on the system time), or generate a random number to use as your random seed.

Also, you should read the manual.

2.5 How do I change the random seed?

To change the random seed to 12345, for example, you can use:

% admixture -s 12345 myFile.ped 3


or if you want the random seed to be generated from the current time, use

% admixture -s time myFile.ped 3

0
Entering edit mode
The deal is, a lot of papers mention that they use different random seeds for each run. Based on my background with BEAST, I thought the program would always give a different random seed. I thought it would do that by default