PLINK: Whole genome data analysis toolset

Warnings & Known Issues

Development of PLINK is ongoing: as such, there is always likely to be a list of features that are only partialy implemented, or have problems of some kind. Our policy is to release a web-based warning for these as soon as any problems are identified, and, for more major problems, to release a patched version on the main download page as soon as possible.

Note that any new patched verions will still give the web-based warning message (i.e. that is based solely on version number, rather than patch).

Want to ignore a warning and run the analysis anyway? Then run PLINK with the --noweb flag to disable the web-check. Not recommended...

Often, the issues flagged here are fixed in the development source code, available from , which you will need to compile yourself.

Format issue with --hardy option

Date: Sep 1 2008

Version: 1.04

Problem: The output of the --hardy command has the fields in the wrong order, with the TEST and A1, A2 fields swapped. Also, the A2 field always equal the A1. This is fixed in development source code available online; will be fixed in next release

Problem with --chap and missing genotype data

Date: Oct 20 2008

Version: 1.04

Problem: A serious problem emerged with the conditional haplotype tests: when there is a reasonable amount of missing genotype data, the p-values calculated can be very liberal. This is because individuals were included under the null model likelihood calculations but not the alternate model calculations under these scenarios. Fixed in development source code; will be fixed in next release. As an approximate, temporary fix, add the option --hap-miss 2 (this will mean that all individuals are phased and included in the test, no matter how much missing genotype data they have)

This document last modified