Dairy Genomics Program of
the Centre for Tropical
Livestock Genetics and Health
Tropical Dairy Genomics
Karen Marshall, ILRI
Eileen Wall, Scotland's Rural College
CTLGH Annual Meeting, Edinburgh, 25-28
September 2017

Program Aims
The dairy genomics program of CTLGH aims to facilitate the
application of genomics to dairy production in the tropics,
for increased livestock productivity, enhanced livelihoods of the
rural poor, increased food and nutritional security, and a more
sustained environment.

Key program focus areas will be:
 Identifying key applications of genomics to dairy
production in the tropics and advocating on these
 Supporting the development of tools and
methodological approaches to facilitate the above
identified applications
 Capacity building
 Partnering in research and resource mobilisation
Program focus area
Initial focus is cattle in Africa – but expect to expand to other
species / regions in later years (e.g. Buffalo; Dairy goats)

Mapping CTLGH Dairy Genomics projects to the CTLGH-BMGF-DFID results framework
Outputs BMGF original funding DFID funding
(* proposed)
Next funding priority
1.1.1 Define and characterize
cattle adaptability or
productivity traits
Dairy cattle adaptation to
(1) protracted nutritional
deficits and (2) disease
1.1.2 Identify putative causal
variants linked to cattle
adaptability/resilience or
productivity traits
Genomic variants for milk
and other ADGG traits*
Signatures of selection
for adaptation e.g. heat
1.1.3 Develop genomic tools
and/or prediction algorithms to
aid the selection of superior
dairy cattle – including in
collaboration with ADGG
Enhanced SNP-chips for
African cattle genome
interrogation and
assignment of breed
composition
Genomics Reference
Resource for African Cattle
Strengthening and testing
wider applicability of ADGG
developed genomic tools
Suitability index for
screening global dairy cattle
for use in East Africa
1.1.4 Identify cattle adaptability
or productivity genetic/
metagenomic variants for use
as proof of concept re gene
editing
Literature study /
new variants arising from
program studies
P3,P4,P5
P3
P4
ADGG
ADGG
ADGG
ADGG

Enhanced SNP-chips for
African cattle genome
interrogation and assignment
of breed composition
Eileen Wall
Karen Marshall
Funded BMGF

…also known as…
Dynamic pipeline for identification
and integration of informative SNPs
to support genetic improvement of
African cattle
Enhanced SNP-chips for African cattle

Intermediate Outcome 1.1: Two proof of concept (POC) genomic/
metagenomic based technologies to aid the selection of superior
locally adapted breeding bulls for use in Africa.
 Bioinformatics pipeline to combine ranges of sequence and SNP
data for analysis
 Integrate alignment of sequence data (output from Project 1.1)
 Set up to deal with multiple SNP chip versions and/or densities to
help deal (associated projects)
Outcomes/outputs

Output 1.1.3: Develop genomic tools and/or prediction algorithms to
aid the selection of superior dairy cattle, in collaboration with the
African Dairy Genetics Gains (ADGG) program.
 By analysing new (and “banked”) sequence and SNP data on African
cattle we will quantify how informative the current range of the
SNP chips are for
 quantification of breed make-up
 prediction of genomic breeding values.
 By doing this will be quantifying the most informative range of SNPs
for the different scenarios of population structures in African cattle
as one tool may not fit all.
Outcomes/outputs

To-date sequence data
 1000bulls.org sequence data integrated into the SNP and sequence data
mgmt system
 Bos taurus vs Bos indicus vs crossbred?
 In some instances we may wish to source “raw” sequence data and realign
using common protocols
 Sequence information from the Genomics Reference Resource for African
cattle (in generation)
 Alignment pipeline established to support Program1 in Edinburgh
 Field-programmable gate array that allow for hardware-accelerated genome
pipeline algorithms
 Optimised for speed and accuracy of mapping, alignment, sorting and
haplotype variant calling
 Tested with 91 Bos taurus samples in cooperation with UoE – 4X faster with
equivalent accuracy

 Need to analyse data from (pedigree?), SNPs and sequence data to
identify most informative SNPs for different breeding scenarios
 Different scenarios (and sub groupings) may require different optimised
SNP chip
 Virtual vs actual SNP chips
 Incorporation of functional results from CTLGH and wider projects
 Develop a pipeline to help integrate the different sources of data
 Is imputation of sub clusters of the data useful?
 Outcome – information and imputation pipeline that can be
customised for different genetic improvement scenarios
To date – bioinformatics data and pipeline

Imputation dynamic pipeline
I
N
P
U
T
D
A
T
A

Imputation dynamic pipeline
I
N
P
U
T
D
A
T
A
Tested on Bos taurus – Imputed 5,000
bulls (50k) to full sequence in 9.1 hours
(dependent on numbers of sequenced animals included)

Imputation dynamic pipeline
 So far tested with “perfect” data – need to integrate the complexity of
African data
 Links with ADGG and other sequence/SNP partners being discussed
 Testing accuracy of pedigree free/hybrid imputation
 Dedicated bioinformatics post to be hired by UoE to undertake
diversity analysis
 Outcome: Pipeline to identify informative SNPs for different scenarios
 Virtual, actual SNP chips with/without imputation
 Addition of functional information and SNP pre-selection.
 With linked projects testing different SNP scenarios impact on accuracy
of (i) breed identification and (ii) genomic based improvement
programs.

Genomics Reference Resource
for African Cattle
Karen Marshall
Eileen Wall
Funded BMGF & Livestock CRP

The Genomics Reference Resource will comprise:
genomic data on African cattle breeds,
metadata on the sequenced or genotyped animals –
including GPS location at time of sampling,
documented and easily searchable outputs arising
from the use of the data (tools, publications, etc.).
Genomics Reference Resource for African Cattle
It is an initiative between
CTLGH and African partners,
with support from AU-IBAR.

Genomics Reference Resource Web-Portal
Collated summary information on the
Genomics Reference Resource; interrogation
ability
Existing genomic
information on
African cattle
breeds
Newly generated genomic
information on African cattle
breeds – contributed by
African partners with potential
assistance from CTLGH
Public sequence databases
Outputs
from data
use
Schema
Partner collects samples
with CTLGH support–
first round breed
prioritization focusing
on African Bos Taurus
Partner
provides
previously
collected
samples
AU-IBAR supporting
linkages to partners of
their genetics project

State of the resource
Breed type Sub-group Number breeds
Bos Indicus Large East African Zebu 8
Small East African Zebu 3
West African Zebu 2
Bos Taurus Humpless Longhorns 4
Humpless Shorthorns 3
Mixed Commercial composite 1
Recently Derived Breeds 2
Sanga 8
Zenga 2
Total 33
Samples: 33 breeds from 13 African countries
Sequencing: 11 breeds (10 animals per breed) submitted thus far
Bos
indicus,
13
Bos taurus, 7
Sanga /
Zenga,
10
Other, 3
NUMBER OF BREEDS OF
DIFFERENT TYPES

Partner countries to-date

Data collected at sampling
• National partner details
• Enumerator details
• Date and time of sampling
General
• Unique identifier
• Owner – name(s) and gender(s)
• GPS location
• Basic descriptors - breed, sex, age,
• Pictures - front and side
Animal
• Unique barcode
• Sample type and medium
• Intended sample destination
Sample
Training in use of ODK tool
for data collection - Sudan

Documentation
Animal owner
consent forms
National research
permit(s)
National animal
ethics permit(s)
National access and benefit
sharing permit(s) – CTLGH
and national authority on
AnGR ABS provider country
Material transfer
agreement(s) –
CTLGH and national
partner
Sample movement
permits (import,
export)
National access and benefit
sharing permit(s) – CTLGH
and national authority on
AnGR ABS user country
ILRI research
permit(s)
ILRI animal ethics
permit(s)
Collaborative
research agreement –
partner and ILRI
Animal and
sample data

Use of the resource - next efforts
Generation of additional genomic information; web-portal
Customised SNP-chips for African cattle for gene identification studies
that more information than the current high density SNP-chip
 P1 for genomic basic of dairy cattle adaptation e.g. heat;
 P4&P1 for genomic basis of dairy cattle disease tolerance;
 other stakeholders.
Customised SNP-chip for determining the breed composition of
individual cattle; also reference population data for the analysis
 P1 in screening animals for characterisation studies;
 other stakeholders – especially Africa research partners – e.g. in support of in-situ breed
comparison studies;
Allele frequency determination in indigenous cattle populations
 P3 for screening of candidate target alleles and recipient breeds;
 other stakeholders.
Resource will continue to be strengthened as funds allow

Strengthening and testing
applicability of ADGG
developed genomic tools
John Gibson
Raphael Mrode
Karen Marshall
Proposed for funding DFID

Strengthening & testing applicability of ADGG developed genomic tools
Background
 The Africa Dairy Genetic Gains (ADGG) project
is a pilot project working in Ethiopia and
Tanzania that is exploring the methods and
structures that will support sustainable
genetic improvement in East African
smallholder dairy cattle populations
 Developed a reduced assay of 200 SNPs to
determine the breed composition (400 if
parentage verification is included) of East
African cross-breed dairy cattle
 Applications include:
 certification of breed composition of crossbred
bulls for natural mating or AI;
 farmer knowledge of cow breed composition –
at time of purchase, or to inform selection of
appropriate bull breed-types
 parent verification

Strengthening & testing wider applicability of
ADGG developed genomic tools
1) Determine whether the reduced snp assays for
exotic dairy proportion & parentage assignment,
developed in DGEA and ADGG, can be used in other
dairy cattle populations outside East Africa
 via combining data from Ethiopia and Tanzania (ADGG)
with data from Senegal and Malawi.
 More genotypes and lager spectrum of breed :
 Senegal - indigenous breeds (Zebu Gobra and Zebu
Maure), and their crosses with exotic Bos indicus
(Guzerat) and Bos Taurus (Montbeliarde, HF)
 Malawi - Malawi Zebu and various Malawi Zebu x
European dairy breed crosses.
 Deliverable = reduced SNP set(s) for determining breed
composition (as exotic dairy proportion) and parentage
assignment for use in African crossbred dairy populations

Strengthening & testing wider applicability of
ADGG developed genomic tools
2) Determine the accuracy of European and/or
indigenous breed composition to be estimated
from snp data
 i.e. extending form proportion indigenous versus
exotic to proportions of actual breed-type
 DGEA data estimated proportions of various
breeds from cross bred data but there is no
control data to verify these
 using data from a large farm with a variety of crosses,
and pedigree and performance records: this will be
combined with ADGG data for refined breed
recommendations
 Deliverable = reduced snp set that estimate individual
breed proportions with sufficient accuracy for use in
the ADGG genetic improvement program; improved
estimates of performance of crosses to different dairy
breeds, and resulting recommendations for use.

Strengthening & testing wider applicability of
ADGG developed genomic tools
3) Determine whether it is possible to impute
from low density assays to high density of snp
with sufficient accuracy to provide useful
accuracy of EBV in crossbred dairy
populations outside of East Africa
 via combining data from Ethiopia and Tanzania
(ADGG) with data from Senegal and Malawi
 Imputation method is established – based on
the pairwise SNP (co)variance and weighted by
MAF (8k - 91% , 20k ---- 95% to HD based on
DGEA data only
 Deliverable = reduced snp set that will allow
imputation to sufficient number of snp to
construct accurate relationship matrices for use
in African crossbred dairy populations

Suitability index for
screening global dairy cattle
for use in East Africa
John Hickey
Raphael Mrode
Karen Marshall
Proposed for funding DFID

Suitability index for screening the global Holstein population
Background
 The ranking of dairy sires is likely to differ between high input dairy systems
in developed countries and the lower-input dairy systems in Africa (G x E)
 Currently we do not have a tool for screening the global Holstein – or other
dairy breed - populations for their suitability as parents of crossbred animals
in Africa
.
A holstein-friesian bull at the Kenya Animal Genetic
Research Centre. Credit, P. Karaimu, ILRI.
 Such a tool can be developed using an
approach called Reciprocal Recurrent
Genomic Selection (RRGS)
 RRGS involves using information
collected on crossbred animals within a
commercial environment to drive
selection decisions in the purebred
nucleus animals from which they derive.

Proposal, key deliverables
 Using genotype data from ADGG (5000 – 10,000
crossbreeds) and from pure Holsteins, develop
and validate a genomic prediction equation for
a “suitability index” based on the RRGS model
 For a index of traits (aligning to ADGG breeding
objective)
 Deliverable = Validated genomic prediction
equation for a “suitability index”
 Screen genotyped Holstein bulls (potentially
from Genus ABS) using the suitability index
 Deliverable = identified Holstein bulls suitable for
use as parents of crossbred animals in East Africa
– this information shared with industry partners
Suitability index for screening the global Holstein population

Dairy cattle adaptation -
disease
Mark Bronsvoort
Andrea Doeschl-Wilson
Georgios Banos
Raphael Mrode
Okeyo Mwai
John Gibson
Karen Marshall
Proposed for funding DFID

Define 3 disease phenotypes
based on screening older
adult dairy cattle for 3 chronic
persistent infections:
 bovine tuberculosis
 brucellosis
 leptospirosis
Adding disease phenotypes to the ADGG project

 Bovine tuberculosis (bTB) is
an important zoonosis with
~3% of human TB cases zTB
 Prevalence varies
 TZ 9-33% in Intensive farms
 CAM 5-25% depending on test
 Incidence likely to rise with
increased dairy farming
 Animal welfare problem
Adding disease phenotypes to the ADGG project

Adding disease phenotypes to the ADGG project

 Brucellosis is a highly contagious zoonosis
caused by ingestion of unpasteurized milk
or undercooked meat from infected
animals, or close contact with their
secretions.
 Abortions; granulomatous lesions in joints
(hygromas), liver, spleen, genitals
 Varying estimates in East Africa
 TZ ~10.8% (Jiwa 1996)
 KEN ~15% )Kadohira 1997)
 TZ 10.6% (Msana 1986)
 Some reports of increased resistance in
Nelore breed compared to Holstein
 Initial association with Nramp1 gene not
substantiated
Adding disease phenotypes to the ADGG project

 Leptospirosis - many serovars with
many origins
 Cattle L. hardjo most important
 TZ ~30% animals and ~58% herds in
Tanga (Schoonman and Swai 2010)
 Boran a risk factor (OR = 2.7) (Swai et
al. 2004)
 Important zoonosis with 13.4% (95% CI
11.1% to 16.1%) slaughterhouse
workers Western Kenya seropositive
(Cook et al 2017)
Adding disease phenotypes to the ADGG project

Adding disease phenotypes to the ADGG project
 Single sampling of ~4000 adult cattle
enrolled on the ADGG project
 Interferon-gamma
 IDEXX brucella ELISA
 Linodane L. hardjo ELISA
 Store down DNA for low density SNP
 Estimate prevalence of 3 zoonoses in
smallholder dairy populations
 (Quantify breed variation in
susceptibility to infection)

Dairy cattle adaptation -
protracted nutritional deficit
Karen Marshall
Alan Duncan
Ilona Gluecks
Mizeck Chagunda
Proposed for funding DFID

Background
 We need to breed animals fit for a future changed environment
 In tropical livestock systems, we do not have sufficient knowledge on
adaptive capacity of different livestock genotypes, or mechanism
underpinning adaptation, to do this
Direct and indirect impacts of climate change on livestock systems
Thornton PK, Boone RB, Ramirez-Villegas J. 2015. Climate
Change Impact on Livestock. CCAFS Working Paper No. 120.
Dairy cattle adaptation to protracted nutritional deficits

Proposal
 To initiate studies on adaptation of dairy cattle breeds / cross-breeds in East
Africa, including the indigenous breeds and recently popularizing cross-breeds
 Focus traits –
 the sensitivity of milk production to feeding levels
 ability to recover milk production after protracted nutritional deficits
(such as that observed during seasonal dry spells, droughts)
Dairy cattle adaptation to protracted nutritional deficits
Why these traits?
feed is a key determinant of dairy
production in East Africa
seasonal changes in feed quantity
and quality are large
feed variability is likely to increase
with increased climate variability

Modular approach
Key deliverables module 1:
 Renovated livestock research facilities on ILRI’s farm,
facilitating assessments of dairy cattle productivity under
different environmental conditions (including feed levels)
 Understanding the relationship between East African
dairy cattle breed / cross-bred type and
 sensitivity of milk production to feeding level and
quality
 animals’ ability to recover milk production after
periods of protracted nutritional deficit
 Samples stored for DNA analysis towards later GWAS
(as data builds up)
 Design of broader GWAS experiment / future modules -
in consultation with experts
Dairy cattle adaptation to protracted nutritional deficits
G x E interaction
The interaction of genotype
and environment that
produces the phenotype
SRUC Dairy Research Centre

‘Kapiti’ renovations
Individual animal feeders (e.g. HOKO)
Automated milking machines and
accessories
Automatic weight-scale
Weather station
Feed preparation equipment – feed
mixer, chaff cutter
Milk analyser (fat & protein)
Computer + back-up
Proposed livestock research facilitates
ILRI’s Kapiti station:
32,000 acres in semi-arid environment,
currently 2200 cattle and 1600 small ruminants

Proposed experiment (module 1)
Feeding regime
Low
grade
dairy
High
grade
dairy
Normal-Normal
Low-Low
Normal-Low
Low-Normal
Varying low /
normal
Animal trial: 120 days, starting 2
weeks after calving; total n=80
Comprehensive records on cow
(milk, health, fertility – return to
heat / pregnancy) and calf
(growth, health)
Identify and purchase high and
low-grade dairy cattle
Low = 75% Zebu +
25% improved dairy
High = 25% Zebu +
75% improved dairy
Artificially inseminate

Genomic variants for milk
and other ADGG traits
John Gibson
Raphael Mrode
Karen Marshall
Next funding priority

Objectives
Identify genomic regions associated with productivity and
functional traits - milk yield, number of times animals are
treated, reproductive performance and longevity
Incorporation of GWAS results to GEBV computation to
optimize productivity and functionality (within ADGG)
Optimise productivity and functional traits

Requirements
Nature of the data (crossbred animals) implies dense chips
(HD) is required to capture LD between markers and QTL
Adequate data size
ADGG genotyping budget is limited - both the number of
animals assayed and the size of the assay used. This will
lower the power of the GWAS
Increase number of animals genotyped
More animals with HD to increase accuracy of imputation
for animals genotyped with lower density
Optimise productivity and functional traits

Approach:
 GWAS on all data set
 Animals of the same breed
composition but with marked
differences in performance
Deliverables:
 Genomic regions associated with
productivity and functionality
 Methods to incorporate these into
GEBV predictions
Optimise productivity and functional traits

Stakeholder engagement
and impact pathways
Karen Marshall

1st stakeholder engagement – “African Cattle Genomics Exchange”
Stakeholder engagement

 Facilitated web-discussion over 2 week period
 Using genomic information on cattle in Africa
 Building the Genomics Reference Resource
 Aims:
 Program awareness and stakeholder input
 Identification of potential partners
 Participation
 Advertised via DAD-Net, BecA network etc.
 427 people registered: 63 countries (29 African)
 190 discussion comments
 52 respondents completed the survey
 10 respondents volunteered samples (15 breeds)
 Discussion summarises at http://cattle-genomix.net/
African cattle genomics exchange - discussion
“Current genomic tools offer a ray
of hope but the basic
infrastructure for conventional
animal performance recording are
still necessary”
“ The design (of the Genomics
Reference Resource) must lead to
exciting and useful results
generated from the early
investment, to provide incentives
for future investments”

Communication products

Other planned initiatives
Identifying key applications of
genomics to dairy production
in the tropics
 engagement process of
stakeholders and experts
Capacity building
Potential for
impact
Technical
feasibility
Will first assess
international interest
Facilitating tropical dairy genomics
research for development (R4D)
through development of an
international consortium

Theory of change & importance of partnerships
Example from Livestock CRP
Genetics Flagship