Relating multivariate shapes to genescapes using phenotype-biological process associations for craniofacial shape
- Department of Cell Biology & Anatomy, Alberta Children’s Hospital Research Institute and McCaig Bone and Joint Institute, Cumming School of Medicine, University of Calgary, Canada
- School of Dentistry, Faculty of Medicine and Dentistry, University of Alberta, Canada
- Department of Biology, Loyola University Chicago, United States
- The Jackson Laboratory, United States
- Department of Medical Genetics, Faculty of Medicine and Dentistry, University of Alberta, Canada
- Department of Orthopaedic Surgery, School of Medicine, University of California, San Francisco, United States
- Department of Animal Biology, University of Illinois Urbana Champaign, United States
Figures
Figure 1
Process multivariate genotype-phenotype (MGP) schematic.
Once a process is selected, we cross-reference the known gene locations using Ensembl with the locations of the genotyped markers in the Diversity Outbred (DO) sample. The founder probabilities of …
Figure 2 with 2 supplements
Process multivariate genotype-phenotype (MGP) for chondrocyte differentiation with a regularization parameter of 0.075.
(A) PLS1 genetic loadings are shown for each gene in the model sorted from largest to smallest effects. Individual founder allele effect sizes are colored within each bar. The gene in red text …
Figure 2—figure supplement 1
10-fold cross-validation results for the chondrocyte differentiation multivariate genotype-phenotype (MGP).
The black-labeled Y axis on the left shows the vector correlation between the process MGP model and a Bmpr1b mutant. The red-labeled Y axis on the right shows the root mean squared error (RMSE) of …
Figure 2—figure supplement 2
Permutation of marker sets of fixed size.
The permuted R2 distribution of 10,000 chondrocyte differentiation multivariate genotype-phenotype (MGP) analyses is shown in blue. The estimated R2 of the chondrocyte differentiation MGP is shown …
Figure 3 with 1 supplement
Chondrocyte defects in Bmpr1b mutants.
(A, B) Quantification of cell size in the sections of the intersphenoid synchondrosis shows an increase in relative cell size as well as a change in the distribution of cell sizes throughout the …
Figure 3—figure supplement 1
Chondrocyte morphometric example.
Landmarks are placed in the top, bottom, left, and right sides of the cell to best capture the height and width of the cells (show here as crosses). The height and width measurements are then used …
Figure 4 with 2 supplements
Process multivariate genotype-phenotype (MGP) for determination of left/right symmetry with a regularization parameter of 0.04.
(A) PLS1 genetic loadings are shown for each gene in the model sorted from largest to smallest effects. Individual founder allele effect sizes are colored within each bar. The gene in red text …
Figure 4—figure supplement 1
10-fold cross-validation results for the chondrocyte differentiation multivariate genotype-phenotype (MGP).
The black-labeled Y axis on the left shows the vector correlation between the process MGP model and a Fgf10 homozygous mutant. The red-labeled Y axis on the right shows the root mean squared error …
Figure 4—figure supplement 2
The permuted R2 distribution of 10,000 L/R symmetry multivariate genotype-phenotype (MGP) analyses is shown in blue.
The estimated R2 of L/R symmetry MGP is shown as a black vertical line.
Figure 5 with 2 supplements
Process multivariate genotype-phenotype (MGP) for palate development.
(A) PLS1 genetic loadings are shown for each gene in the model sorted from largest to smallest effects. Individual founder allele effect sizes are colored within each bar. The gene in red text …
Figure 5—figure supplement 1
10-fold cross-validation results for the palate development multivariate genotype-phenotype (MGP).
The black-labeled Y axis on the left shows the vector correlation between the process MGP model and a Ankrd11 homozygous mutant. The red-labeled Y axis on the right shows the root mean squared error …
Figure 5—figure supplement 2
The permuted R2 distribution of 10,000 palate development multivariate genotype-phenotype (MGP) analyses is shown in blue.
The estimated R2 of palate development MGP is shown as a black vertical line.
Figure 6 with 1 supplement
Gene drop tests.
For each of the example analyses, we show the effect of removing the most heavily loaded markers from the process multivariate genotype-phenotype (MGP) analysis on the (A) variance explained by the …
Figure 6—figure supplement 1
Single marker importance.
For a sample of 500 processes, we show the proportion of processes with a single marker with 0.5×, 2×, and 3× higher effect size than any other marker. 51.6% of processes show a marker effect 0.5× …
Figure 7 with 1 supplement
Pairwise multivariate genotype-phenotype (MGP) vector correlations.
(A) Pairwise correlations of phenotypic effects for 15 process MGP analyses. Scale on the right denotes color correspondences to vector correlation, where yellows are high correlations, greens are …
Figure 7—figure supplement 1
Pairwise process multivariate genotype-phenotype (MGP) vector correlations as a function of the number of shared genes between the processes.
The pairwise vector comparisons have been subset to only include the processes shown in Figure 7A, with four pairs highlighted. Processes that share less than 10 genes can produce very similar and …
Figure 8
Comparisons of multivariate genotype-phenotype (MGP) and mouse mutant directions.
(A) Seven MGP phenotypes projected onto a principal component analysis (PCA) of the Diversity Outbred (DO) and a sample of 30 mutant mouse genotypes. Mutant means are labeled in black. The …
Figure 9 with 1 supplement
Example screenshot of web version of process analysis.
Analyses include a barplot of the relative effect sizes of each selected marker and the associated phenotype shown with black vectors at each landmark. If a mutant comparison is selected, the vector …
Figure 9—figure supplement 1
Combining queries in the multivariate genotype-phenotype (MGP) shiny app with the pipe operator.
In order to filter the Gene Ontology (GO) database with multiple terms, the pipe operator can be used as shown. Here, the user has selected processes associated with either the apoptosis or Wnt …
Figure 10 with 1 supplement
54 3D landmark configuration.
(A) Sagittal view of representative scan with landmarks shown as red spheres. (B) Dorsal view of landmark configuration. (C) Ventral view of landmark configuration.
Figure 10—figure supplement 1
Demographic plots for the Diversity Outbred (DO) sample.
(A) The distribution of the sample by generation and data source (lab). (B) Distribution of sex by source (lab).
Author response image 1
