SEARCH

SEARCH BY CITATION

The documentation of the algorithm presented at the Appendix of the original article “Robustness of Bayesian Multilocus Association Models to Cryptic Relatedness” by Hanni P Kärkkäinen and Mikko J Sillanpää, published on Volume 76 Issue 6 (November 2012) of the Annals of Human Genetics at pages 510–523, contains the following clerical errors.

The GEM-algorithm for the EBL, step 5 (bottom right, p. 13):

The expected value of inline image should be

  • display math

The GEM-algorithm for the EBL, step 6 (top left, p. 14):

Update the hyperparameter values of δ and inline image to their conditional expectations, the conditional expectations should be

  • display math
  • display math

In the GEM-algorithm for the Indicator model, step 4 (top right, p. 14) the error variance is called inline image instead of inline image, and the text should be: Update the error variance inline image into its conditional expectation.

The GEM-algorithm for the Indicator model, step 6 (middle right, p. 14) the formula for computing inline image should be

  • display math
  • display math

As the errors occurred only in the written text, not in programming, all of the results presented in the paper are based on a correct algorithm. The authors deeply regret that this has happened and are sincerely grateful to Prof. Hiroyoshi Iwata for drawing the errors to their attention.