P resent at ion Tit le G oes Here …presentation subtitle. “Next” generation of decision support tools Guy Davenport CIMMYT Leader IRRI-CIMMYT Crop Research Informatics Laboratory Jayashree B. Bioinformatics ICRISAT
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 1/26
Present at ion T i t le Goes Here
…presentation subtitle.
“Next” generation of decision supporttools
Guy Davenport
CIMMYT Leader
IRRI-CIMMYT Crop Research Informatics LaboratoryJayashree B.
Bioinformatics
ICRISAT
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 3/26
:: color, composition, and layout
What is decision support in breeding?
• Decide breeding strategy
o Decisions on mating designs, population sizes,phenotyping and genotyping strategies, selectionmethods and intensities
• Choice of parental lineso Integration, analysis and presentation of all
available information on potential parents withrespect to the target traits and environments.
• Recurrent selection
o Selection and advancement decisions based
relevant genotypic and phenotypic information
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 4/26
:: color, composition, and layout
ParentalLines
Phenotype/Genotype
Cross/Self/Inter-mate
Selection
Make a selectionbased on phenotype
or genotype
NewLines
Parental line
selection
Where do we need it?
Devise breeding
Strategy
Simulation / Prediction
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 5/26
:: color, composition, and layout
Why do we need decision support inbreeding?
• Increased speed in making decisions
o Quick turnover essential especially in multi-seasonbreed programs
• More/better information for making decisionso More markers means better decisions, but
increases complexity
• Better analysiso New analyses need to be made user friendly and
integrated with existing tools and data
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 7/26
:: color, composition, and layout
Methodology and tools are available toassist breeders
• Analysis of the data such as marker effects orselection indices
• Visualization tools relying on combination of sortingand filtering.
• These tools range from
o standalone tools for visualization that oftencarryout a limited range of tasks
o to command-line statistical applications that havea great deal of flexibly but require seriousprogramming knowledge.
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 8/26
:: color, composition, and layout
Problems with current support• Phenotypic, genotypic and genetic data required for
these tools are usually available from disparatesources and formats
• Are not integrated with databases
• Labour intensiveo Requires a lot of reformatting of data from one
format to another
o Not user friendlyo Requires statistics / programming knowledge
• Does not support large numbers of markers
• Software not fully open/free!
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 9/26
:: color, composition, and layout
MB Support Tool - Development plans
• A single desktop application that will consist of
o Molecular Breeding Design (MBD) Module toprovide support for design of initial crosses andstrategy
o Molecular Selection (MoSel) Module to providesupport for subsequent selections
• Will use the extended GCP informatics platform toretrieve both breeding and public data from local andweb resources and create new germplasm lists
• Integrate with simulation tools
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 10/26
:: color, composition, and layout
Molecular Breeding Design (MBD) Module
• Select parental germplasm based on
o presence or absence of traits in potential parents
o compatibility between donors and recipients
o genetic distance matrices based on pedigree(COP) and where available molecular data
o analysis of background similarity via geneticdistance matrices
• Design target genotype and create crossinglists/input files for Molecular Breeding Platform
• In development at ICRISAT
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 11/26
:: color, composition, and layout
MBD Module – Current functionality (1)• Read germplasm lists,
genotype data, trait data, map
data• Display graphical genotypesof potential donors &recipients
• Display linkage maps andQTL
• Check availability ofgenotyping data for selectedmarkers
• Generate list of unscreenedmarkers, file to be passed tomolecular LIMS
Graphical
genotypes
Linkagemap
QTLs
Unscreened
markers list
Export files
to LIMS
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 13/26
:: color, composition, and layout
MBD – User feedback
Demonstrated tobreeders atICRISAT. Furtherdevelopment willfollow closely twoMAB projectsDonor
Recurrent
Target
Distance
matrix
Allele drag and
drop from donor to
target
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 14/26
:: color, composition, and layout
MBD Module – Future functionality• Overlaying of QTL data: multiple populations, multiple
environments, multiple traits.
• Provide for integration and display of additional trait metadata
• Input file formats for QTL data – currently PlabQTL (iMAS), toadd QTL cartographer besides a generic QTL data template.
• Allow flanking markers to multiple QTLs to be selected and sortfor multiple markers before a choice of suitable recurrent ismade.
• Allow display of consensus linkage maps and positioning ofQTLs by marker name rather than map position.
• Expanded tool tip to include color coded information on additiveeffects and LOD scores of individual QTLs.
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 15/26
:: color, composition, and layout
Molecular Selection (MoSel) Module
• Facilitate the selection of the most promising lines in terms of
closeness to the target genotype and likelihood of reaching thetarget according to a proposed development strategy.
o Integrate and display genotype and phenotype information
o Order, group and filter lines• Based on selection indices, marker associations
o Send data to simulation tool to predict how the program maycontinue
o Choose lines and crossing schemes for further development
• Send to MB informatics platform (MB)
• In development at CIMMYT
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 16/26
:: color, composition, and layout
MoSel – Current functionality
Query integrate anddisplay genotypeand phenotype
information
Filter and sort
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 17/26
:: color, composition, and layout
MoSel Module – Future functionality (1)
• Partitioned display of foreground and background
markers• Integration with R statistical software to assign and
useo
marker effects for different loci.o Selection Indices• Extendable so that newly developed analytical
measures can quickly integrated and tested
• Filter, sort and scroll genotype displays for test linesby proximity to target and recurrent lines based ongenetic and phenotypic values
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 18/26
:: color, composition, and layout
MoSel Module – Future functionality (2)
• Aid selection of lines and send back to the MB
platform as crossing lists for the next cycle• Markers fixed in recipient populations should also be
identified so that they do not need to be processedfor subsequent generations.
• Highlight suitable polymorphic markers for use in thedefinition of marker-trait associationso For example identification of x number of evenly
distributed polymorphic markers for mapping
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 20/26
:: color, composition, and layout
GIDs must bepresent in ICIS forall parental lines
Record Phenotype
and Genotype data
in ICIS
Generate phenotypeor genotypeworkbooks
Generate derivativeGIDs in ICIS / MFB
Initial genotypingpolymorphisms
detected
Parental
Lines
Phenotype
/GenotypeSelf/Cross
Selection
Molecular Selection(MoSel)
New
Lines
Molecular BreedingDesign (MBD)
Simulation ToolsQuLine / QuMARS
etc
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 22/26
:: color, composition, and layout
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 23/26
:: color, composition, and layout
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 24/26
:: color, composition, and layout
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 25/26
:: color, composition, and layout
Latest version on DTMA wiki http://wiki.cimmyt.org/confluence/display/DTMAunder Molecular Breeding
8/9/2019 MBP Davenport
http://slidepdf.com/reader/full/mbp-davenport 26/26
:: color, composition, and layout
AcknowledgmentsDevelopment Team
• Luis Avila, Martin Senger, Richard Bruskiewich, Jose Crossa(IRRI-CIMMYT)
• Jayashree B, Prathyusha C, Reetu T (ICRISAT)
• Graham McLaren (GCP)
Proposed User Teams
• Gary Atlin, Yunbi Xu (CIMMYT) and Sarah Hearne (IITA) -
Maize• Yann Mannes and Susanne Dreisigacker (CIMMYT) - Wheat
• Tom Hash, Nepolean T (ICRISAT) – Pearl millet
• Pooran Gaur (ICRISAT) - Chickpea