Riview
Journal
"
Accounting for relatedness in family-based association studies:
application to Genetic Analysis Workshop 18 data
"
Name of Journal
|
BMC
Proceedings
|
Title of Journal
|
Accounting for
relatedness in family-based
association
studies: application to Genetic
Analysis
Workshop 18 data
|
Date of Journal
|
13-17
October 2012
|
Methods
|
phenotype data analysis using real and simulated phenotype
first set. This analysis is done without the knowledge base of the simulation
model. Genotype data were cleaned using standard procedures. This resulted in
four individuals were excluded because of insufficient numbers of their data.
|
The Main Discussion
|
A number of
different methods/software packages have been proposed in the last few years
that implement linear mixed-model approaches to account for population
structure and relatedness among samples in genome-wide association studies
(GWAS), but no detailed comparisons among them have been made before our
effort. Indeed, when a new method/package is developed, it is often quite unclear
whether or how it differs substantially from those already available. To
address this question, we explored the performance of various implementations
of such methods in the longitudinal Genetic Analysis Workshop 18 (GAW18) data
set.
|
Summary
|
All
methods performed well and results were similar, particularly at the most
significant SNPs. We conclude that (at least for nonlongitudinal traits) it
makes little difference to the results which method/software package is used,
and the user can make the choice of package on
the
basis of personal taste, speed, or computational convenience. For
longitudinal traits (modeled without regard to their longitudinal nature) the
slight differences seen between the methods would be an interesting topic for
further investigation, but it is beyond the scope of the current article.
|
Suggestion
|
more
emphasis on improving methods,
especially for traits elongated (models without
regard to the nature of longitudinal)
|