GIBBS2F90

Summary

Ignacy Misztal, University of Georgia, 3/7/2001
Gibbs2f90 is a modification of gibbs1f90 to sample correlated effects jointly. This modification results in better mixing for models with correlated effects such as maternal or random-regression models. Memory requirements and CPU time per round are somewhat higher than in gibbs1f90.
See PREGSF90 with genotypes (SNP) for options.

Running

You can run as

gibbs2f90 parameter_file --rounds 1000000 --burnin 10000 --thin 10 --thinprint 10

in which case you give the number of rounds, the burn-in, the storage of samples in the file gibbs_samples, and the interval between printouts in the screen. Otherwise you can simply type gibbs1f90 and you will be asked the parameter file and the questions below:

number of samples and length of burn-in?

In the first run, if you have no idea about the number of samples and burn-in, just type your guess (10000 or whatever) for samples and (0) for burn-in. You may need 2 or 3 runs to figure out the convergence.

Give n to store every n-th sample?

Gibbs samples are usually highly correlated, so you do not have to keep all samples. Maybe every 10th,20th, 50th, …

To check the convergence and to calculate posterior means and SD, run postgibbsf90.

OPTION fixed_var all 1 2 3

All solutions and posterior means and SD for effects for effects1, 2, and 3 are stored in “all_solutions” and in “final_solutions” every round using fixed variances. Without numbers, all solutions for all effects are stored.

OPTION fixed_var mean 1 2 3

Posterior means and SD for effects1, 2, and 3 in “final_solutions”.

OPTION solution all 1 2 3

Caution: this option will create a huge output solution file when you run many rounds and/or use a large model. All solutions and posterior means and SD for effects1, 2, and 3 are stored in “all_solutions” and in “final_solutions” every round. Without numbers, all solutions for all effects are stored.

OPTION solution mean 1 2 3

Caution: this option will create a huge output solution file when you run many rounds and/or use a large model. Posterior means and SD for effects1, 2, and 3 in “final_solutions”.

OPTION cont 10000

10000 is the number of samples run previously when restarting the program from the last run. When using “OPTION cont”, all output files will be replaced by new ones. Before running with this option, all files should be backed up.

OPTION prior 5 2 -1 5

The (co)variance priors are specified in the parameter file.
Degree of belief for all random effects should be specified using the following structure:
OPTION prior eff1 db1 eff2 db2 … effn dbn -1 dbres
effx correspond to the effect number and dbx to the degree of belief for this random effect, -1 corresponds to the degree of belief of the residual variance.
In this example 2 is the degree of belief for the 5th effect, and 5 is the degree of belief for the residual.

OPTION seed 123 321

Two seeds for a random number generator can be specified.

OPTION SNP_file snp

Specify the SNP file name to use genotype data.