RESEARCH ARTICLE Open Access
Genomic scan of selective sweeps in thin and fat
tail sheep breeds for identifying of candidate
regions associated with fat deposition
Mohammad Hossein Moradi1*, Ardeshir Nejati-Javaremi1, Mohammad Moradi-Shahrbabak1, Ken G Dodds2 and
John C McEwan2
Abstract
Background: Identification of genomic regions that have been targets of selection for phenotypic traits is one of
the most important and challenging areas of research in animal genetics. However, currently there are relatively
few genomic regions identified that have been subject to positive selection. In this study, a genome-wide scan
using ~50,000 Single Nucleotide Polymorphisms (SNPs) was performed in an attempt to identify genomic regions
associated with fat deposition in fat-tail breeds. This trait and its modification are very important in those countries
grazing these breeds.
Results: Two independent experiments using either Iranian or Ovine HapMap genotyping data contrasted thin and
fat tail breeds. Population differentiation using FST in Iranian thin and fat tail breeds revealed seven genomic
regions. Almost all of these regions overlapped with QTLs that had previously been identified as affecting fat and
carcass yield traits in beef and dairy cattle. Study of selection sweep signatures using FST in thin and fat tail breeds
sampled from the Ovine HapMap project confirmed three of these regions located on Chromosomes 5, 7 and X.
We found increased homozygosity in these regions in favour of fat tail breeds on chromosome 5 and X and in
favour of thin tail breeds on chromosome 7.
Conclusions: In this study, we were able to identify three novel regions associated with fat deposition in thin and
fat tail sheep breeds. Two of these were associated with an increase of homozygosity in the fat tail breeds which
would be consistent with selection for mutations affecting fat tail size several thousand years after domestication.
Background
The domestication of livestock represents a crucial step
in human history. The rise of civilizations could not
happen without domestication of plants and animals.
Sheep (Ovis aries) is the first grazing animal known to
have been domesticated [1]. Multiple mitochondrial
lineages suggest that domestication occurred several
times, as in other livestock species such as cattle, goat
and pig [2]. Recognition of the origin of domestication
is difficult by the fact that the first domestic animals
were no different from their wild counterparts [3]. In
spite of this, the archaeozoological evidence suggests
that the domestication of sheep occurred during the
Neolithic revolution approximately 9000 years ago [4] in
a region in northern Iraq and nearby regions in Iran [5].
Since domestication, sheep have established in a wide
geographical range due to their adaptability to poor
nutrition diets, tolerance to extreme climatic conditions
and their manageable size [6].
Fat tail breeds are an important class of sheep breeds
that are first documented as being present 5000 years
ago. The earliest depiction of a fat tail sheep is on an
Uruk III stone vessel of 3000 BC and fat and thin tail
sheep appear together on a mosaic standard from Ur
dated around 2400 BC [3,7]. The fact that the fat tail
breeds are now prevalent in the Fertile Crescent, where
sheep were originally domesticated, while thin tail sheep
breeds are predominant in peripheral areas [7] and that
the wild ancestor of sheep is thin tail suggest that the
* Correspondence: Hoseinmoradi@ut.ac.ir
1Department of Animal Science-Excellent centre for improving sheep carcass
quality and quantity, University of Tehran, PO Box 3158711167-4111, Karaj,
Iran
Full list of author information is available at the end of the article
Moradi et al. BMC Genetics 2012, 13:10
http://www.biomedcentral.com/1471-2156/13/10
© 2012 Moradi et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
许瑞霞
高亮
n. 驯养;教化
first domesticated sheep were thin tail and fat tail was
developed later. The evidence shows that sheep were
being farmed throughout Europe 5000 years ago [8].
The fat tail is considered as an adaptive response of
animals to a hazardous environment and is a valuable
energy reserve for the animal during migration and win-
ter. Until recently it had additional value to the herder
because it was used to preserve cooked meat for longer
periods of time and also as an energy reserve during
times of drought and famine. Therefore the climatic var-
iation as well as the associated requirements of humans
led to artificial selection for higher fat tail weight across
generations [9-11]. Nowadays in intensive and semi-
intensive systems most of the advantages of a large fat
tail have reduced their importance and therefore, a
decrease in fat tail size is often desirable for producers
and consumers. Fat deposition requires more energy
than the deposition of lean tissue, animal fat has lost
much of its market demand and monetary value and
sheep producers have easy access to other forms of aux-
iliary feeding [12]. These breeds are commonly found in
a wide range of countries in Asia especially the Middle
East and North Africa [7]. To date, several investigations
into the inheritance of fat tails have been undertaken
[10,11,13-18], nevertheless the genes affecting fat deposi-
tion in fat tail breeds are still unknown.
The study of genes underlying phenotypic variation
can be performed in two different ways, firstly from
phenotype to genome, which is performed by LD based
association mapping or by targeting particular candidate
genes identified based on homology to known genes,
and secondly from genome to phenotype, which involves
the statistical evaluation of genomic data to identify
likely targets of past selection using selective sweep ana-
lysis [19,20]. The elimination of standing variation in
regions linked to a recently fixed beneficial mutation is
known as a “selective sweep” and has recently been the
focus of much theoretical and empirical attention
[21-23].
In contrast to natural populations, domesticated spe-
cies provide an exciting opportunity to understand how
artificial selection promotes rapid phenotypic evolution
[22]. With an hypothesis that different selection pres-
sures operated in thin and fat tail breeds over the his-
tory of time and somehow the selection acts on a
variant that is advantageous only in one breed, it is
expected that the frequency of that variant may differ
across populations to a greater extent than predicted for
variants evolving neutrally in all populations [21]. Identi-
fying of these genome regions, which have been subject
to such selective sweeps could reveal the mutations
which are responsible for the fat deposition in these
breeds. The examination of variation in SNP allele fre-
quencies between populations, which can be quantified
by the statistic FST, is a promising strategy for detecting
signatures of selection [21,24].
To date a relatively small number of examples have
successfully identified genomic regions subject to posi-
tive selection in different domestic animals [20,25-38].
The constraint to identifying selection signatures in
sheep has been the limited density of markers. This lim-
itation has recently been reduced with the availability of
tens of thousands of single nucleotide polymorphisms
(SNPs) using the Ovine SNP50k BeadChip (http://www.
sheephapmap.org.) In this research two independent
resources have been investigated to identify regions
associated with fat deposition in these breeds. The first
data set was comprised of one thin and one fat tail Ira-
nian sheep breed while the second was comprised of
several other fat and thin tail breeds selected from the
Ovine HapMap project.
Sheep production constitutes the most important
component of the Iranian livestock industry with a total
of approximately 50 million head. Twenty seven breeds
and ecotypes have been documented in Iran, the major-
ity derived in situ. The Lori-Bakhtiari sheep breed is
one of the most common indigenous breeds. It is well
adapted to the hilly and mountainous Bakhtiari region
stretched out to southern Zagros Mountains (Figure 1),
with a population of more than 1.7 million. The animals
are kept mostly in villages under semi-intensive systems.
Relative to other Iranian fat tail breeds the Lori-Bakh-
tiari is a large breed, having the largest fat tail by girth
and weight [39]. Sheep breeds vary in tail length and
this breed, perhaps surprisingly, has a rather short tail
length. Zel is the only thin-tail Iranian breed and it is
present largely on the northern slopes of the Elburz
mountain range near the Caspian Sea [40] representing
around 3% of Iranian sheep.
The aim of this study was to find selective sweeps
between the Iranian thin and fat tail breeds using dense
SNP markers and to compare with those from similarly
divergent breeds extracted from the Ovine HapMap
dataset (http://www.sheephapmap.org.) Our work pro-
vides the first genome wide characterization of selective
sweeps in thin and fat tail breeds. Each animal was gen-
otyped with approximately 50,000 SNPs and a variety of
selection sweep tests were utilized.
Results
Genotyping and data mining
A total of 94 animals consisting of 47 samples per breed
were genotyped on the Illumina OvineSNP50K Beadchip
assay in the Zel-Lori Bakhtiari data set. One animal in
each breed had greater than 5% missing data and these
were excluded from further analysis. The second data
set used in this study was SNP genotyping data from
similarly divergent breeds in the Ovine HapMap project
Moradi et al. BMC Genetics 2012, 13:10
http://www.biomedcentral.com/1471-2156/13/10
Page 2 of 15
(Table 1). After data cleaning (see methods) a total of
45,611 and 48,053 SNPs passed the filtering criteria in
the Zel-Lori Bakhtiari and HapMap data sets respec-
tively and were included in the final analyses. The over-
all average distances between 2 adjacent SNPs were
relatively consistent among the chromosomes, being
about 60 kb in the Zel-Lori Bakhtiari and 58 kb in the
HapMap data sets. In all cases the locations used were
obtained from OAR true chromosomes (ver.1.0, as at 5/
2008) from CSIRO [41]. The principal component ana-
lysis (PCA) in Zel-Lori Bakhtiari data set, using the
individual SNPs as the data, resulted in the first two
principal components (PC1 and PC2) explaining 18.6%
and 2.9% of the variance respectively. We found that
PC1 separated out the two breeds from each other while
one animal in each breed was separated from all other
animals for PC2 (Figure 2). These two animals were
excluded from further analysis. Finally, 45 animals (36
females and 9 males) per breed passed the data cleaning
steps and 95% and 95.1% of all remaining SNPs were in
Hardy-Weinberg equilibrium at the 5% level in Zel and
Lori-Bakhtiari breeds respectively. For the SNPs
Figure 1 Traditional distributions of the two Iranian breeds used in this study are shown in (a) with breed examples of Lori Bakhtiari
(b) and Zel (c).
Moradi et al. BMC Genetics 2012, 13:10
http://www.biomedcentral.com/1471-2156/13/10
Page 3 of 15
analyzed in this study, the average MAF over all samples
was 0.29 (SD = 0.13) in the Zel-Lori Bakhtiari data set
and 0.30 (SD = 0.12) in the HapMap data set.
Genomic distribution of FST in Zel-Lori Bakhtiari data set
The plot of windowed FST against location is shown in
Figure 3. The average of differentiation between Zel and
Lori Bakhtiari breeds was 0.024 (SD = 0.035). As shown
in this figure, in several instances outlier SNPs tended
to cluster to similar regions. Specifically, we found evi-
dence of selection in seven regions with windowed FST
value > 0.20 on chromosomes 2 (between 55,861-56,300
kb), 2 (between 73,631-73,784 kb), 3 (between 146,615-
146,676 kb), 5 (between 47,149-47,263 kb), 7 (between
30,512-30,585 kb), 7 (between 46,642-46,843 kb) and X
(between 58,621-61,452 kb).
The average FST for autosomal and X-linked SNPs was
significantly different (0.024 and 0.035, respectively; t
test, t = 6.2, P < 10-10). A higher average FST for X-chro-
mosome SNPs could occur because of its smaller effec-
tive population size compared with that of the
autosomes.
Another genetic distance measurement including
unbiased estimates of FST as described by Weir and
Cockerham [42] was also calculated. Because the results
were highly correlated (r = 0.995) with the above results,
so have not been presented.
Study of Bovine genes and published QTLs in regions
showing evidence of selection
Seven regions showing the largest signals of selection in
Zel-Lori Bakhtiari data set were chosen for further ana-
lysis. As the current annotation of the sheep genome is
not as comprehensive as cattle, the regions of interest in
O. aries were compared to the corresponding areas in B.
taurus. Dot plots for corresponding areas of Ovine and
Bovine genomes showed strong co-linear relationships
between the two considered sequences in all regions
and rearrangements were not observed (Additional file
1: Figure S1).
A summary of orthologous area in both species and
published bovine genes and QTLs is presented in table
2. Orthologus genes in the bovine genome were identi-
fied using the BLAT genome search with UCSC Gen-
ome Browser [43].
Online databases of published QTL in beef and dairy
cattle, show that the regions identified here were pre-
viously been found to be in regions harboring QTL
affecting fat and also carcass yield traits (Table 2). For
example, the regions on chromosomes 2 overlapped
with QTLs previously suggested as being related with fat
depth, and also chromosomes 2 (second region) and 5
with fat thickness and both regions on chromosomes 7
with milk fat yield.
Table 1 Origin and sample size of different breeds included in this study
Data set Breed Code1 Tail status Sampling area Sample size
Zel-Lori Bakhtiari data set
Zel ZEL Thin tail Iran 47
Lori-Bakhtiari LOR Fat tail Iran 47
Ovine HapMap data set
Deccani DEC Thin tail India 23
Scottish Black Face SBF Thin tail United Kingdom 56
Bunder Oberland Sheep BOS Thin tail Switzerland 24
Gulf Coast Native GCN Thin tail North America 94
Karakas KAR Fat tail Turkey 18
Norduz NDZ Fat tail Turkey 20
Afshari AFS Fat tail Iran 37
1 Code is the 3-letter code for each breed.
Figure 2 Animals clustered on the basis of principal
components analysis using individual genotypes. Zel and Lori
Bakhtiari breeds are shown by green and blue circles respectively.
Moradi et al. BMC Genetics 2012, 13:10
http://www.biomedcentral.com/1471-2156/13/10
Page 4 of 15
Genomic distribution of FST in Ovine HapMap data set
and comparison with Zel-Lori Bakhtiari data set results
An independent resource comprised of similarly diver-
gent breeds for thin and fat tail breeds in the Ovine
HapMap project were selected to determine whether the
regions with large allele differentiation in Zel-Lori Bakh-
tiari data set could be confirmed. The pattern of FST
across the genome was also calculated in the Ovine
HapMap data set (fat and thin tail breeds were pooled)
and values were averaged in a sliding window. The
windowed FST was then plotted against location in the
genome. The windowed FST values, at each common
SNP, were correlated in the two data sets with r =
0.413, N = 45,238 and the average of differentiation
between thin and fat tail breeds in the Ovine HapMap
data set was 0.027 (SD = 0.038). The sliding window
average FST revealed good agreement between both data
sets for regions on chromosomes 5, 7 (second region)
and X (Figure 4).
Calculation of empirical p-values for windowed FST
values in each data set (Table 3) shows that the 3 regions
with the largest differentiation in Zel-Lori Bakhtiari data
set are also significant in the Ovine HapMap data set (P
< 0.01). On the other hand, combining p-values from
independent tests of significance in Zel-Lori Bakhtiari
and Ovine Hapmap data sets using Fisher’s combined
test revealed the first region on chromosome 2 (P <
0.0001) in addition to these same significant regions.
The Weir and Cockerham method of estimating FST
[42] was also performed for the ovine HapMap data and
a high correlation (r = 0.992) was observed. Previous
reports [e.g. [24,52]] also indicated these methods led to
similar results; therefore, a strong correlation between
these two measures is not surprising.
Study of median homozygosity in candidate regions
Median run of homozygosity (Figure 5) was increased at
the candidate regions on chromosome 5 and X for Lori
Bakhtiari (fat tail) and at the candidate region on chromo-
some 7 for Zel (thin tail). The largest differences of med-
ian homozygosity were located on chromosome X and this
was present for a longer distance as well, whereas these
statistics were lower on Chromosome 5. A study of med-
ian homozygosity in the HapMap data set for thin and fat
tail breeds (data not shown) revealed similar results.
Estimation of effective population size and haplotype age
The average extent of LD in the genome was used to
estimate the effective population size at various times in
the past. The estimated effective population size show a
persistent decline from 2000 down to 20 generations
ago, declining from 4900 (both breeds) to 840 in Zel
and 532 in Lori Bakhtiari.
The graph of effective population sizes suggests a dis-
tinctive time point when the breeds separated ~1,100
generations ago (Additional file 2: Figure S2). Assuming
an average generation interval in sheep of around 5
years or 5500 before present, this is congruent with the
first archaeological evidence for the fat tail sheep breeds
(~5,000 years ago).
The average effective population size over the period
was calculated using reciprocals as described by Fal-
coner [53]. Using this average estimated effective popu-
lation size and based on the current frequency of
sweeps in our regions of interest [54], the age of sweeps
under the assumption of selection neutrality and genetic
drift has been inferred for chromosomes 5, 7 and X to
be approximately 7100, 9600, and 6900 generations
before present respectively.
Figure 3 Distribution of windowed FST values for Zel versus Lori-Bakhtiari breeds by chromosome. SNP position in the genome is shown
on the X-axis, and windowed FST is plotted on the Y-axis. Regions with arrows above had windowed FST value > 0.20 and were later examined
for further analysis.
Moradi et al. BMC Genetics 2012, 13:10
http://www.biomedcentral.com/1471-2156/13/10
Page 5 of 15
Table 2 Bovine genes and published QTLs in regions showing evidence of selection in Zel vs Lori Bakhtiari data set
Chr-Region Location on
Ovine genome
Location on
Bovine genome
Gene* RefSeq Number QTL QTL
Reference
2-1 2:55859169-56302905 8:62601056-63034390 NPR2 NM_174126.2 Fat Depth [44]
SPAG8 NM_001102005.1 Carcass weight [45]
HINT2 NM_174340.2 Body weight (birth) [45]
TMEM8B NM_001103335.1
2-2 2:73628609-73861345 8:45226663-45415081 — Fat Depth [44]
Fat thickness [44]
Carcass weight [45]
Body weight (birth and mature) [45]
3 3:146615284-146676235 5:33981017-34041987 CACNB3 NM_174509.3 Meat [46]
Tenderness
DDX23 NM_001102202.1 Hot Carcass Weight [47]
Birth Weight [48,49]
5 5:47146900-47332222 7:44936435-45113988 PPP2CA NM_181031.2 Fat thickness [48]
SKP1 NM_001034781.1
TCF7 NM_001099186.1
7-1 7:30510065-30587784 10:29363930-29441529 —— Milk fat yield [50]
Carcass weight [48]
Hot Carcass Weight [48]
Body weight (birth) [45]
7-2 7:46639859-46910155 10:45312778-45581699 PTGDR NM_001098034.1 Milk fat yield [50]
Hot Carcass Weight [48]
Body depth [51]
X X:59192476-60151772 X:51751347-52712765 —
* Obtained by orthology with the Bovine genome
Figure 4 Distribution of windowed FST values for Zel-Lori Bakhtiari a