the 7th Mexico 2011 1 by K6p19B


									     Feedbacks from Participants to the 7th Workshop on QTL
               Mapping and Breeding simulation

Day 1: Breeding Simulation (4:30pm, 20 January 2011)

Ravi Singh:

Simulation plays important role in improving the efficiency of breeding program.
Simulations of modified pedigree compared to selected bulk strategies,
demonstrated that selected bulk method was slightly better, reconfirming belief
that selected bulk was better.

In the HarvestPlus program, we aim to incorporate exotic germplasm by
backcrossing. With complex traits, simulation provided insights in to the number
of backcrosses required, and proved useful to back-up ideas of breeders.

A third example of simulation saved the Wheat Program a lot of money. There
was a strong push to adopt double haploids in the CIMMYT’s wheat breeding
program, and Hans was pushing to establish experiment to compare DH with
traditional breeding. Ravi suggested that simulation could be used instead of
field breeding experiment. The results of simulation demonstrated that DH was
not better than current breeding methods in the spring wheat program, but was
beneficial in the winter wheat program. Ravi said “best piece of work”, that
“must be published”.

Other areas where simulation may be beneficial:
    Compare what was done 7-8 years ago with present methods, to quantify
       the enhancements achieved.
    Seeds of Discovery Program – requires rapid incorporation of new
       germplasm. Simulation of genomic selection could be used to
       demonstrate potential value.

Ravi said “I am a strong supporter of simulation to evaluate the potential of new

Hans Braun:

Spring wheat program used to make very large number of crosses. But now
spend a lot of money on the characterisation of populations. With increasing
availability and emphasis on molecular breeding we have a number of questions
that need to be addressed:
    How many alleles can be selected at the same time
    Population size
    Can we make this process more efficient

Comments by Hans on simulation of double haploid breeding discussed earlier
by Ravi: spring wheat program is a special case, due to being able to complete
two generations in same year, with large segregating populations. This allows
program to rapidly increase frequency of desirable alleles. The benefits of the
spring wheat program were “difficult to sell without the use of simulation” and
“need to publish paper”.

Jose Crossa:

Application of selection indices was compared with simulation. But need to
repeat across more populations, and to include epistasis. Selection index is more
complex when markers are also included along with multiple traits, multiple
environments, gxe, and differing levels of noise and heritability.

Simulation has proven very useful. QuGene simulation is well proven, and
software is available for independent validation of published results.

Genomics poses new challenges, and believes that model that includes epistasis
will improve selection using markers or genomic scans. Simulation can be used
to demonstrate applicability and validity of models across populations.

David Bonnett:

Has been exposed to genetic simulation over many years, initially with Mark
Cooper. Subsequently became aware of modules that were being developed by
Jiankang Wang. Then started to see the applicability of simulation through
applications to the CIMMYT wheat program, but still had doubts that a breeder
would be able to implement simulation without significant assistance, despite
many discussions with Jiankang Wang and Scott Chapman.

Believes that simulation software is not getting to the point where a breeder may
be able to run themselves. “Sees to be closer” “more breeder friendly” but
support services may be required to help breeders. “You really learn a lot as a
breeder” through discussions and setting-up and running computer simulations
of your breeding program.

Jose Crossa: believes that there will be a direct relationship between usability of
interface and the number of users.

Ravi Singh on cross prediction: The real value will come when you have known
information, and have computer simulations that tells you what crosses to make,
and then generations those crosses by simulation. Such a program should
integrate all available information and then choose the crosses (not the breeder).
Probably will be little use if breeder has to go through crosses one by one.
“Simulation will assist both new breeders and seasoned breeders”.

To top