Title: Believing in MAGIC: Validation of a novel experimental breeding design
1Believing in MAGIC Validation of a novel
experimental breeding design
- Emma Huang, Ph.D.
- Biometrics on the Lake
- December 2, 2009
2Multiparent Advanced Generation Inter Cross
3Experimental crosses
A
B
B
AB
AB
AB
44 Parents
A
C
B
D
AB
CD
ABCD
6 generations of selfing
Inbred individuals
5Final Result
6In theory...
- No genotyping for users after setup is complete
- Effective design for G x E
- Resource for large community - many traits
- Large population of RILs
- Large genetic (allelic) and phenotypic diversity
- Ability to map epistatic interactions
- High recombination ? high resolution
RIL final lines
Size
Diversity
7Vital resource for linkage mapping
- No physical map/sequence for wheat (yet)
- Previous maps developed for specific population
- Limited polymorphisms
- Would have to join maps across populations
- Possibly inconsistent estimates across maps
- Many markers have not been mapped
- MAGIC map is potentially
- More complete due to greater genetic diversity
- More accurate due to larger population size
- More precise due to many generations of
recombination
8But... nontrivial
- Complex inheritance from founders
- Limited genotyping
- Population size/ markers computational burden
- Marker issues dominant markers, polyploidy,
etc.
9Theory vs. Reality
10Linkage Map Construction
- Basic Strategy
- Filter and preprocess marker data
- Estimate pairwise recombination
- Group and order loci
- Refinement
11Step 2 Estimating Recombination Distance
- In standard designs, counting numbers of
different genotypes - This doesnt work for 4/8 way crosses
- Many generations of recombination
- No intermediate genotyping
- Ambiguous inheritance of alleles
- Instead maximize the likelihood
12Step 4 Refinement
- Start with framework map
- Position markers relative to fixed locations
- Maximize likelihood over grid of positions
- Compared to initial ordering
- Iterative and time-consuming
- Less sensitive to missing values
- Additional information about marker relationships
13MAGIC bag of tricks
- R package mpMap
- Simulate data, filter/process, generate linkage
map, visual quality checks
14In a perfect world
- Nice data
- Fully informative markers
- No missing data
- No genotyping errors
Chr 6
Recombination fractions below diagonal scaled
LOD scores above
Chr 5
Chr 4
Chr 3
Chr 2
Chr 1
15Something closer to reality
- Typical data
- Biallelic markers
- 10 missing data
- 10 genotyping errors
- gt datbad lt- mp.sim(map, simped, seed1,
error.prob.1, missing.prob.1) - --------------------------------------------------
----- - Summary of mpcross object
- --------------------------------------------------
----- - 0 markers were removed with missing values in
founders - 0 markers were removed with non-polymorphic
founder genotypes - --------------------------------------------------
----- - 195 markers were biallelic.
- 0 markers were multiallelic.
- --------------------------------------------------
----- - 195 markers had gt5 missing data.
- 99 markers had gt10 missing data.
- 0 markers had gt20 missing data.
- --------------------------------------------------
----- - 49 markers had lt1e-5 p-value for segregation
distortion - 2 markers had lt1e-10 p-value for segregation
distortion - 0 markers had lt1e-15 p-value for segregation
distortion
164-parent MAGIC
- DArT SNP markers
- Constructed map
- 871 progeny
- 1148 markers
- 20/21 chromosomes
- 2010 5000 lines from 8-way cross
17Chromosome 6B
Genetic Map
Heat Map
18Looking to the future
- Improve the current map
- Starting from the least informative set of
markers - Further genotyping to fill in gaps
- QTL Mapping
- Testing different approaches
- Field trials
- Association Mapping
- Using constructed map for complex analysis
19CSIRO Mathematical and Information Sciences Emma
Huang Research Scientist Phone 61 7 3214
2953 Email Emma.Huang_at_csiro.au
Thanks to Andrew George Colin Cavanagh Matthew
Morell
Thank you
Contact UsPhone 1300 363 400 or 61 3 9545
2176Email Enquiries_at_csiro.au Web www.csiro.au