Genotyping the DNA of dairy cattle has now become routine practise by A.I. companies, breeders
wishing to merchandise genetics and many producers looking for an additional tool to manage their
herd of cows and heifers. The adoption of DNA genotyping services offered through Holstein Canada
has expanded from the initial attraction of elite breeders and A.I. companies to a widespread interest
due to the introduction of lower density testing panels since September 2010. At that time, not only did
the cost of genotyping with the initial 50K panel (including roughly 50,000 markers) drop from $250 to
the current $140, but a new “Low Density” panel with 3,000 markers was launched at a genotyping cost
of $47 per animal. More recently, however, the latest, new and improved version of a low density (LD)
panel was released, which doubled the number of markers to 6,000 but did not result in any change of
genotyping costs. Other than this seemingly “more for your money” argument from a genotyping
perspective, how does this translate into more accurate genomic evaluations and genetic selection
decisions?
Genotyping Activity
Table 1 presents the current number of dairy cattle genotypes that exist at Canadian Dairy Network
(CDN) for use within each breed in Canada. These include all genotypes from across North America as
well as those received via the genotyping of foreign animals and/or the exchange of genotypes with
other countries such as Italy and United Kingdom in Holsteins. Genotyping activity varies significantly
across breeds due to population size as well as the gradual development and implementation of
genomic evaluations over time. In Canada, genomic evaluations were first official in August 2009 for
Holstein and this was followed in April 2010 for Jersey and August 2011 for Brown Swiss. Preliminary,
unofficial genomic evaluations are currently being calculated by CDN for Ayrshire and genotyping
services are still very limited for Guernsey and other breeds. The Canadian research effort towards
genomic evaluations in breeds with smaller population size has focused on the use of “High Density”
(HD) panels with roughly 800,000 markers, which explains the proportionally high counts for HD
genotypes in Ayrshire and Guernsey (Table 1).
Table 1: Number of Genotypes by Panel Type Within Breed (January 2012)
Genotyping Date First Used Holstein Jersey BrownSwiss Ayrshire Guernsey
Panel in Canada
50K 2008 81,394 6,765 2,123 129 0
3K-LD Sept. 2010 48,073 9,532 473 3 5
6K-LD Oct. 2011 10,230 1,384 64 6 0
HD May 2011 1,017 9 12 481 81
In Canada, all DNA samples submitted for 3K genotyping processed through Holstein Canada since
late October 2011 have been genotyped using the new 6K Low Density panel. The genotyping request
form for the GenoTest program, available on the CDN and Holstein Canada web sites, has therefore
been updated for the user to select “Low Density (3K/6K) Chip”, which will automatically yield a 6K
genotype.
Implications for Genomic Evaluation
To-date, all genomic evaluation systems used in any country worldwide are based on 50K genotypes,
meaning they have roughly 50,000 markers. For animals actually genotyped with the 50K panel, these
genotypes are the automatic outcome from the laboratory analysis. Animals submitted for HD
genotyping receive a genotype with roughly 800,000 markers but this can easily be reduced to a 50K
genotype based on the exact subset of markers included on the 50K panel. Low Density (LD)
genotypes resulting from the use of the 3K or 6K panels include a subset of markers from the 50K
genotype but the remaining marker genotypes are missing. The ultimate success of using 3K
genotypes available since September 2010 has been dependant upon the development of accurate
methods for deducing or “imputing” the full 50K genotype from the subset of 3,000 markers available on
the 3K panel. Research in this area has progressed extremely fast around the world and, in Canada,
has been led by Drs. Mehdi Sargolzaei and Jarmila Johnston through ongoing collaboration between
the Centre for Genetic Improvement of Livestock (CGIL) at the University of Guelph, L’Alliance Boviteq
and CDN. With the recent surge in use of the 6K genotyping panel in Canada, this same team of
researchers have aimed to quantify the gain in accuracy achieved from the 6K genotypes compared to
the low density 3K panel previously used.
Accuracy of Imputation
There are various factors that affect the degree of accuracy by which 50K genotypes can be imputed
from lower density ones. While one of these is genotyping errors at the laboratory level another one is
the marker density of the starting genotype, which in this case are either 3,000 or 6,000 markers spread
across the entire genome. The higher the number of markers used in the 6K panel gives a higher
marker density and therefore a greater accuracy of imputation. Imputation accuracy is also a function of
the size of the genotyped reference population available and the methodology applied. At CDN, the
method used, which maximizes the accuracy for as many animals as possible, combines the strengths
of family-based imputation and population-based imputation.
Table 2: Decrease in Imputation Error Rate Using 6K vs 3K Panel for Genotyping - Holstein
Sire Genotype Dam Genotype Average Imputation Error Rate
3K Panel 6K Panel
50K 50K 0.9% 0.4%
50K 3K 1.8% 1.1%
50K 6K 1.4% 0.7%
50K Not Genotyped 2.9% 1.1%
Not Genotyped Not Genotyped 5.9% 2.3%
Not Genotyped Not Known 7.7% 5.0%
Table 2 shows the decrease in imputation error rate for 6K genotypes compared to 3K genotypes in
Holsteins. In each case outlined, the 6K panel led to imputed 50K genotypes that were more accurate.
When the animal’s sire is genotyped with the 50K panel, which covers approximately 90% of all cases
in the Holstein breed, the average imputation error rate is one percent or less. As expected, the
imputation accuracy reduces when the sire is not genotyped and especially when the dam is not known
in the pedigree record. In these cases, while the 6K panel shows clear advantages over the previous
3K panel, the imputed genotyped used for genomic evaluation will have a lower accuracy than if the
animal was directly genotyped with the 50K panel.
Summary
The genotyping services offered by Holstein Canada to dairy cattle producers of various breeds in
Canada concentrate on the standard 50K panel or the “Low Density” (LD) panel. Originally, a 3K panel
was used for low density genotyping but since October 2011 this was automatically replaced in Canada
with the new 6K panel. A “High Density” panel is also available for genotyping but its current use has
been limited to research projects and for genotyping key reference sires in each dairy breed. The good
news for dairy producers of any breed is the increased imputation accuracy provided by the 6K panel
compared to the former 3K panel and there has been no increase in genotyping costs associated with
the transition to this improved technology.
|
