1) The document outlines research using genomics tools like GBS and DArT analysis to advance yam breeding for traits like yield and quality. Over 900 yam accessions were analyzed to identify genetic diversity and population structure. 2) Highly polymorphic SNP markers were validated and used in GWAS to identify genomic regions associated with traits. Bi-parental populations were also analyzed to identify QTLs. 3) Molecular markers were developed for early prediction of flower sex, an important trait for yam breeding. The research aims to apply markers to accelerate yam breeding through genomic selection.

1. ’Striving for
excellence in yam
breeding using
genomics tools’

2. Agre Paterne, David De Koeyer, Ranjana Bhattacharjee,
Guillaume Beauchet, Andreas Gisel, Yusuf Muyideen, Edemodu
Alex, Asrat Amele, Adrian Powell, Ryo Matsumoto, Sato
Muranaka, Patrick Adebola, Kohtaro Iseki, Ryochi Terauchi,
Susan Strickler, Kolade Olufisayo, Lukas Mueller, Lava Kumar,
Robert Asiedu

3. Outlines
 Introduction
 GBS analysis
 DArT Analysis
 SNP selection and validation
 LGC data analysis and comparison of platforms
 Genome wide association study (GWAS)
 QTL identification using bi-parental population
 Application of Molecular Markers
 Ongoing activities and conclusion

4. Cultivated species
• D. rotundata
• D. alata
• D. cayenensis
• D. dumetorum
Used for food, economical, and
for social aspects
Breeding program for gene
introgression
 Yam food crop & cash crop in West Africa
 600 species
Wild and related
• D. praehensilis
• D. bulbifera
• D. togoensis
• D. mangenotiana
Brief introduction

5.  Vegetative propagation
 Heterozygous
 Multiple ploidy levels (2x, 3x, 4x…)
 Poor flowering / seed set
 Low multiplication rate​
 Multiple uses/tuber used both as food and propagule​
 Serious pest, disease and storage problems
 Long breeding cycle through conventional scheme
 Little information about the genetic diversity among and
between species
 No reference clone to address the mixture issues
Brief introduction
Challenges in yam breeding

6.  Increased yam productivity
 Genetic improvement of two species
 Breeding for yield and for other traits
 Establishing a breeders’ community
 Phenotyping for bi-parental
population mapping and genome-
wide association studies for key
agronomic and quality traits
 Create capacity building among the
breeding programs
AfricaYam Project goals
How can we use genomic tools to
address these challenges and also
to achieve these goals ?????

7. ❑947 accession were analyzed using GBS methods (7 species belong to several
groups).
❑GATK and Tassel software were used to analyze the raw data, plink and vcf tools were
used to call the final SNP markers
❑ Quality control following summary statistics, IBS methods was used to identify
duplicate accessions, DAPC and HC were used for clustering
Species Number Groups
D. rotundata 892 Cultivated, breeding lines, landraces & market
D. cayenensis 23 Semi wild
D. praehensislis 7 wild
D. togoensis 6 wild
D. mangenotiana 5 wild
D. burkilina 7 wild
D. abyssinica 2 Semi wild
Material and methods
GBS analysis

8. TASSEL
all SNP
GATK
all SNP
TASSEL
filtered
GATK
filtered
TASSEL
maf 005
TASSEL
maf 001
GATK
maf 005
GATK
maf 001
Comparing TASSEL and GATK SNP based
on physical position and allele variant: raw
vcf vs filtered vcf (raw (“all”) vs 80% missing
data SNP removed (“filtered”)
Comparing TASSEL and GATK filtered
SNP based on physical position and
allele variant(maf >0.01 and maf 0.05)
SNP comparison using two pipelines
GBS analysis

9. Duplicates
Identification of duplicate accessions through
IBS and population structure
414 accessions as unique clones
Cultivated
wild

10. Identification of duplicate accessions
through IBS and population structure
Population well
structured
through the HC
and admixture at
K = 5
Using K number of clusters equal
to 5, the total clones were grouped
into different group based on the
genotype profile:
Wild species were grouped in the
same group while cultivated
(landraces and breeding) were
distributed in different groups

11. Species
PCA and relation among the species
GBS data was able to give
clear representation of the
population structure through
PCA. Species were grouped
based on there respective
origin and evolution ui
Dioscorea rotundata (TDr)
Dioscorea abyssinica
Dioscorea burkilliana
Dioscorea praehensilis
Dioscorea togoensis
Dioscorea cayennsis
Dioscorea mangenotiana

12. Phylogeny and relationship among
the species
GBS data filtered at MAF > 0.05
were able to classify the different
species involved in this study
with high accuracy using FST
and cultivated species were
classified in the same group with
low genetic diversity.
Phylogeny tree using Fst: TDb: Dioscorea burkilia; TDt: Dioscorea togoensis; TDc:
Dioscorea coyenensis, TDr: Dioscorea rotundata; TDa: Dioscorea abysinica; TDm:
Dioscorea mangitiana; TDp: Dioscorea prahensilis

13. 94 clones from both D.alata and D. rotundata include landraces
and breeding clones
Current Ref Genome
size (bp) without
chr00
456,674,974
+ 143.7 MB not
assigned
SNP density with
59000 SNP
DArT data analysis

14. Dioscorea rotundata Breeding lines
Dioscorea alata Breeding lines
Dioscorea rotundata Landraces
Dioscorea alata Landraces
DArT data was able to provide clear distribution of the two species. However,
two clones with rotundata genotype were located in alata cluster
DArT data analysis

15. Selection of highly polymorphic SNPs
and primer validation from GBS data

16. SNP control quality for SNP design
LD
with r
Marker
retained
MAF>0.1 PCA Fst
pairwise
r2=0.9 3266
r2=0.8 2891
r2=0.7 2545
r2=0.6 2315
r2=0.5 2121
r2=0.4 1916
r2=0.3 1683
r2=0.2 1415
r2=0.1 1032 700 400 322
plink --file data --indep-pairwise 10 3 r2
4117 SNPs used for the LD pruning
loop_fetch_segment.pl
blastn
SNP with MAF > 0.1

17. Control quality for SNP design
IUPAC position in the data base was also considered for the final selection
and led to 192 selected SNPs which were used for PCA and HC

18. PCA analysis using the 192
selected SNP
HC analysis using the 192 selected SNP
from the total set
Selection of highly polymorphic SNPs
and primer validation

19. LGC validation and quality
Selection material
for SNP validation
Number Raison
GBS materials (TDr) 115 Known genotype
TDa 12 Test
Interspecific 22 Test
TDd 1 Test
TDr from different
trial
120 Test
TDc 5 Semi-wild
TDpr 10 Wild
Total 285
Failled
49%
Worked
51%
LGC VALIDATION
3%
50%
47%
SNP qualities
Bad
Monomorph
polymorph
285 clones were selected for the primer validation
including different species and putative
interspecific hybrids

20. LGC validation and quality using DAPC
DPCA using 192 SNP DPCA using 97 SNP
DPCA using 46 SNP
DAPC analysis also confirm the representativeness of the selected SNP and
that these can be used in breeding program for many purposes

21. LGC results
Out of the 192 SNP selected for primer validation on 285 clones, 97
SNPs successfully amplified and produced results.
The 97 SNP markers were able
to make a clear distinction of the
different species involved in the
primer validation.
However, the interspecific
clones involved in this study did
not show any difference from
the D. rotundata parents.

22. Whole genome scanning for key traits in
yam
318 Dioscorea rotundata landraces and breeding lines were used for
genotyping using whole genome resequencing. Clones were planted in
augmented design in 2018
Morphologic and agronomic
canopy architecture
female flowering intensity
first inflorescence initiation time
plant sex estimation
Stem diameter computation
Average tuber length measurement
Spines on stem
Average tuber weight
Average tuber width
Dry matter content
Tuber shape estimation
Tuber surface texture
Tuber thorniness intensity
flowering intensity estimation
inflorescence type
stem color
plant vigor estimation
leaf apex shape
stems per plant
Flesh oxidation color
Hairiness of tuber
Time to 50% foliage
senescence
Total tuber weight per plant
Leave density
Leave shape estimation
Abiotic and biotic
virus severity estimation
anthracnose disease severity estimation

23. Whole genome scanning for key traits in
yam
 R was used for summary statistics and to normalize the data; SAS
was used to estimate the broad sense heritability.
 BLUP predicted values were generated for all the traits
 Plink was used for the admixture analysis and the CV was linked
to the BLUP for the association mapping alongside with the
genotype file
 Kinship matrix was developed and a Manhattan plots were
generated using qqman script in R and Tassel
 Regions link to the each trait were then identified using LOD
scores

24. Whole genome scanning for key traits in yam
Modified Hisat2 script, Picard and GATK were
used for SNP calling
Raw data sequencing was aligned with the
pseudomolecules and the percentage of alignment
was calculated for each clone
Minimum % of reads mapped: 36.96%
Average % of reads mapped: 68.28%
Maximum % of reads mapped: 79.07%
Total SNPs identified: 14,905,577
 WGR was used to genotype the clones in IBRC

25. Whole genome scanning for key traits in
yam
Traits with normal distribution
Traits with abnormal distribution

26. Whole genome scanning for key traits in
yam
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
H2

27. Whole genome scanning for key traits in
yam

28. Whole genome scanning for key traits in
yam
Chr1
6%
Chr2
8%
Chr3
3%
Chr4
5%
Chr5
7%
Chr6
9%
Chr7
4%
Chr8
7%Chr9
4%
Chr10
4%
Chr11
4%
Chr12
4%
Chr13
8%
Chr14
4%
Chr15
2%
Chr16
7%
Chr17
5%
Chr18
5%
Chr19
2%
Chr20
2%
Chr21
1%
SNP NUMBER

29. Whole genome scanning for key traits in
yam
Farmer’s varieties
Breeding lines
Gene bank accessions

30. Whole genome scanning for key traits in
yam

31. Whole genome scanning for key traits in
yam

32. Morphologic and agronomic
canopy architecture
female flowering intensity
first inflorescence initiation time
plant sex estimation
Stem diameter computation
Average tuber length
measurement
Spines on stem
Average tuber weight
Average tuber width
Dry matter content
Tuber shape estimation
Tuber surface texture
Tuber thorniness intensity
 Two populations(D. alata & D.
rotundata) were involved in the QTL
identification
Leaves were collected from 202 clones
for TDa and 166 TDr were sent to CIRAD
for genotyping
27 traits were selected from the yam crop
ontology for phenotyping
Phenotyping data were collected on single
plant basis from experiment planted in an
augmented design
flowering intensity
estimation
inflorescence type
stem color
plant vigor estimation
leaf apex shape
stems per plant
Flesh oxidation color
Hairiness of tuber
Time to 50% foliage
senescence
Total tuber weight per
plant
Leave density
Leave shape estimation
Abiotic and biotic
virus severity estimation
anthracnose disease severity estimation
Population
Phenotyping
Variables
Mapping population analysis for QTL
identification

33. Mapping population analysis for QTL
identification
Progress on genotyping of the two population for QTL identification
 DNA was extracted from the two
population with good quality
 GBS libraries have been
constructed
 First result of the fastq file has
been received
 SNPs will be called and associated
with phenotype data for QTL study
and targeted primer design.
 Phenotyping data will be collected
for another season and QTL will be
validated next season
DNA qualities of some clones

34. Molecular markers for flower sex prediction in yam at early
stage

35. D-actin
D-actin
D-actin
SP16
SP16
SP16
High rate for PCR amplification for
both SP16 and D-actin
36%
64%
Sex distribution
Male
Female
Sex distribution using Marker
Molecular markers for flower sex prediction in
yam at early stage
Plant Material:
190 clones were selected from different stage of breeding program:
seedling nursery (unknown sex) and other plants from advanced stage
(sex known and unknown)

36. Clone
Sex in filed
2017
D-actin
amplification SP16 Amplification Genotype
Sex based
on primer
Prediction
accuracy %
TDrAkunchi Female 1 1 ZW Female 100
TDr89/02475 Female 1 1 ZW Female 100
TDr89/02665 Female 1 1 ZW Female 100
TDrNduu Female 1 1 ZW Female 100
TDr11/00873 Female 1 1 ZW Female 100
TDr95/19156 Female 1 1 ZW Female 100
TDr97/00793 Female 1 1 ZW Female 100
TDrAlumaco Male 1 1 ZW Female 0
TDrFaketsa Male 1 0 ZZ Male 100
TDr11/00163 Monoecious 1 0 ZZ Male 100
TDr95/01932 Male 1 0 ZZ Male 100
TDr05/00491 Mixture 1 1 ZW Female 100
Next Action
Field sex assessment of the 190 clones using visual scoring for proper
correlation
Validate the result and implement MAS application in the yam breeding
program
High correlation was
observed between
the phenotype and
marker results
Molecular markers for flower sex prediction in
yam at early stage

37. Ongoing activities and conclusion
 Reasonable progress have been made on yam genomics, but a lot still
needs to be done​
 Application of MAS has been implemented but still need to be extended to
other traits
 Finalizing HTPG marker selection process for routine activities in yam
breeding program
 Validating region controlling traits with the second year GWAS
phenotyping data
 Finalizing QTL analysis on the bi-parental populations for D. alata and D.
rotundata using SNP markers
 Comprehensive markers linked to key traits in yam will be developed
 Establish yam genome hub for the breeding community

38. Acknowledgments
 Yam Breeding Unit &
AfricaYam project
Kwabena Darkwa
Chidinma Nwachukwu
 Melaku Gedil
 Abberton Michael
 Rabbi Ismail
 Bunmi Olasanmi
 Surya Saha
 Bioscience Unit
 Unachukwu, Nnanna
 Adewumi Adeyinka
 Obi Queen
 JHI Scotland Dundee
 David Marshall
 BTI Cornell University
 CIRAD France Montpelier
 IBRC Japan
 JIRCAS
Merci beaucoup pour
votre attention