Genome-wide genetic association of complex traits
in heterogeneous stock mice
William Valdar
1
, Leah C Solberg
1,4
, Dominique Gauguier
1
, Stephanie Burnett
1
, Paul Klenerman
2
,
William O Cookson
1
, Martin S Taylor
1
, J Nicholas P Rawlins
3
, Richard Mott
1
& Jonathan Flint
1
Difficulties in fine-mapping quantitative trait loci (QTLs) are a major impediment to progress in the molecular dissection of
complex traits in mice. Here we show that genome-wide high-resolution mapping of multiple phenotypes can be achieved using
a stock of genetically heterogeneous mice. We developed a conservative and robust bootstrap analysis to map 843 QTLs with an
average 95% confidence interval of 2.8 Mb. The QTLs contribute to variation in 97 traits, including models of human disease
(asthma, type 2 diabetes mellitus, obesity and anxiety) as well as immunological, biochemical and hematological phenotypes.
The genetic architecture of almost all phenotypes was complex, with many loci each contributing a small proportion to the total
variance. Our data set, freely available at http://gscan.well.ox.ac.uk, provides an entry point to the functional characterization
of genes involved in many complex traits.
The mouse is a key model organism for understanding gene function
in mammals, yet many mouse phenotypes of interest to biomedical
research have poorly understood and complex, polygenic origins.
Despite new genomic resources such as access to dense maps of
sequence variation and the ability to interrogate the expression levels
of virtually every gene, molecular dissection of the loci that contribute
to quantitative variation remains a challenge
1
. A central problem that
impedes the cloning of QTLs is the difficulty of resolving genetic effects
into sufficiently small intervals to make gene identification possible.
Successful strategies for high-resolution mapping ideally should be
able to identify small genetic effects for any phenotype across the
entire mouse genome. The classical approach begins by genetic
mapping in a cross between two inbred strains or, more recently, in
chromosome substitution strains
2
, and it eventually results in the
identification of a small number of independently segregating loci
mapped into intervals larger than 20 Mb. Subsequent fine-mapping
typically proceeds by repeatedly backcrossing one inbred strain onto
another to isolate each locus. Such attempts are frequently frustrated
when it is discovered that a single QTL segregating in inbred crosses
fractionates into multiple smaller effects, each of which typically
contributes less than 5% to the total phenotypic variance
1
.
We have developed alternative methods for fine-mapping small-
effect QTLs that use outbred mice of known ancestry
3–5
. By exploiting
historical recombinants that have accumulated in a genetically hetero-
geneous stock of mice descended from eight inbred progenitor strains
(A/J, AKR/J, BALBc/J, CBA/J, C3H/HeJ, C57BL/6J, DBA/2J and LP/J)
6
,
we have shown that QTLs explaining 5% or less of the phenotypic
variation can be mapped into intervals of o1 cM. Although the
heterogeneous stock has been used thus far for fine-mapping single-
QTL intervals
4,7,8
, it is theoretically ideal for high-resolution mapping
of multiple QTLs across the genome: its derivation from multiple
founders means it should contain more QTLs than any inbred cross,
and the use of pseudorandom breeding for over 50 generations (in the
case of the stock discussed in this paper) should result in an average
distance between recombinants of o2 cM.
However, a number of potential obstacles must be tackled before
the heterogeneous stock becomes a tool for genome-wide QTL
mapping. First, the high costs of performing whole-genome associa-
tion in the heterogeneous stock may preclude its use, because B100
times more markers and ten times more animals are required
compared with an inbred strain cross
1
. Second, random fluctuations
in allele frequencies and unrecognized selective pressures operating
during the production and maintenance of the stock could reduce its
heterozygosity, with consequent reductions in QTL resolving power.
Third, a whole-genome analysis in the heterogeneous stock, which
involves the simultaneous identification of multiple QTLs, poses
unknown analytical problems that could seriously vitiate the outcome.
The extensive repertoire of methods developed for whole-genome
analysis of a classical intercross
9–12
are not directly applicable: in a
heterogeneous stock, more loci are tested than individuals, so it is not
possible to use methods that fit all markers simultaneously, and many
more parameters are estimated at each locus than in a classical cross
4
.
In this paper we demonstrate the utility of the heterogeneous stock
for high-resolution whole-genome association analyses of quantitative
Received 18 April; accepted 13 June; published online 9 July 2006; doi:10.1038/ng1840
1
Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive, Oxford OX3 7BN, UK.
2
Peter Medawar Building for Pathogen Research,
Nuffield Department of Medicine, University of Oxford, Oxford OX1 3SY, UK.
3
Department of Experimental Psychology, University of Oxford, Oxford OX1 3UD, UK.
4
Current address: Medical College of Wisconsin, Human and Molecular Genetics Center (HMGC), 8701 Watertown Plank Road, Milwaukee, Wisconsin 53226, USA.
Correspondence should be addressed to J.F. (jf@well.ox.ac.uk) or R.M. (rmott@well.ox.ac.uk).
NATURE GENETICS VOLUME 38 [ NUMBER 8 [ AUGUST 2006 879
ARTICLES
© 2006 Nature Publishing Group http://www.nature.com/naturegenetics