SIMIBD example - Handout from 10/18/2000 course

The following is a file prepared partly by Richa Agarwala describing a simple use of SimIBD on a real data set. I have removed references to the data set to avoid revealing any research information. This describes a cookbook usage.

The reference for SimIBD is:

Davis S, Schroeder M, Goldin LR, Weeks DE (1996) Nonparametric simulation-based statistics for detecting linkage in general pedigrees. Am J Hum Genet 58:867-880.

Summary

Do not use SimIBD version 1.x.y as it has been superseded by version 2.1 (and possibly later versions).

Required input files:

• locusXYZ:LINKAGE/FASTLINK format locus file
• pedigree file: LINKAGE/FASTLINK format pedigree file

One output file is produced.

SimIBD currenlty does only 2-point analysis.

SimIBD produces two output files:

• simibd.out
• simibd.aff.out

Steps:

1. simibd pedigree file locus file
   We are only doing SimIBD computation and not SimAPM or SimKIN.
2. Choose weighting function from
   • A : f(p) = 1
   • B : f(p) = 1/sqrt(p)
   • C : f(p) = 1/p
   
   Choose option B; this is strongly recommended in the documentation.
3. Specify number of replicates; a recommended value is in brackets, but you must specify some number
4. Specify number of BOOTSTRAPS per SIMULATED NULL replicate. Again, a number is recommended in brackets.
5. Specify number of SIMULATED NULL DISTRIBUTION replicates total. Again a number is recommended in brackets.
6. Trait locus number: 1
   This is the LINKAGE convention.
7. Enter the marker locus number:

This creates simibd.out which summarizes the results of the runs and simibd.aff.out which is the full output from affected comparisons.

The important table is simibd.out with headings: Family ZObs # Aff. Weight Weighted ZObs PValue Range

This will be followed by 1 row of results for each pedigree in your data set.

The Weight column here has to do with the number of affected people in the family, and is unrelated to the allele frequency weight that you were asked to select. For this program you should literally look at the "bottom line" and see if the p-value is small. p-values < 0.05 are considered significant.

SimIBD can be run in parallel. Contact tran@helix.nih.gov for more information.