B4EST short course: multiomics prediction
Abdou Wade, Harold Durufl?? and Leopoldo Sanchez
INRAE Biofora : June 23, 2022
Global context
What do we mean by genetic architecture?
Architecture implies interacting elements. How important and pervasive are interactions?
1 Multi-omic data might be the clue to inferences of interactions and architectures
2 Multi-omic data might be the clue to inferences of interactions and architectures
Simply, what are the methods for multiomics integration?
Concatenation method: advantages and disadvantages
1 Experimental setup
2 Experimental setup
3 Experimental setup
4 Experimental setup
Results: prediction accuracy
Results: eQTL
Results: average change in ranking over CIS, TRANS, non-eQTL
Results: average change in ranking VS advantage of concatenation
Results: distribution of changes in effects
Workflow
Loading and processing the input data
Load phenotypic and genotypic data
Compute Input distance matrix (241 x 241)
Genomic relationship matrix calculated as proposed by (VanRaden, 2008)
Trancriptomic variance-covariance matrix
Predictions Models
Setting Cross validations folds Sampling
Single omic models
With genotypic data as predictors
With transcriptomic data as predictors
Multi-omics models
Concatenation of both omics data (c_GTBLUP model)
Transformation based integration (t_GTBLUP model)
Plot Predictions output
Global context
What do we mean by genetic architecture?
slide 1
Architecture implies interacting elements. How important and pervasive are interactions?