Morphometric Analyses of Shape: The Analysis Software Toolbox for Craniofacial Shape Quantification in Zebrafish

Our research uses morphometric software to quantify whether ethanol and bmp7a interact to cause subtle craniofacial defects in zebrafish. Linear measures miss subtle changes. With this method, we use landmarks to quantify shape changes while controlling for size differences.

To begin, open the tpsDigs2 software. Click Input source, then File, and select the saved TPS Notepad file. Click Options to visualize the Image tools option.

Set the reference length to 100 micrometers and press on Set scale. Click OK to confirm the scale parameters, then exit the Image tools window. Then, click the aim symbol and place the first landmark at the midline joint between the Meckel's cartilages.

Place the second landmarks at the bilateral joints between Meckel's and the palatoquadrate cartilages. Place the third landmark at the midline joint between the ceratohyal cartilages and the fourth landmark at the bilateral joints between the palatoquadrate and ceratohyal cartilages. Then, place the fifth landmarks at the distal end of the hyomandibular cartilages, ensuring landmarks are placed in the same order for every image.

After placing landmarks, click File to open the dropdown menu. Select Save data, and then Overwrite to save the updated file. Exit the tpsDigs2 software.

Launch the MorphoJ software. Click File, select Create New Dataset, and assign a name to the dataset. Click TPS and select the Notepad file containing the newly added data points.

Then, create the dataset. In Project Tree, click the dataset. Select Preliminaries, then New Procrustes Fit.

Choose Align by principal axes and click Perform Procrustes Fit. Under Preliminaries, select Generate Covariance Matrix to obtain procrustes coordinates. Execute the function when prompted.

Now, select Create or Edit Wireframe and link the points on the images. Click linked points and accept or create the image, then edit classifiers as needed. Open a spreadsheet program while keeping MorphoJ open.

Enter classifier information such as GENOTYPE, TREATMENT, and GENOTYPE TREATMENT, Then, save the file as a CSV file. In MorphoJ, click File and select Import Classifier Variables to import the CSV file. Return to Project Tree and click the dataset.

Under Preliminaries, select Edit Classifiers to confirm all images are included. In Project Tree, select CovMatrix, then click Variation at the top of the page. Select Principal Component Analysis to calculate principal component scores.

Click PC scores to display the generated graph. Right-click on the graph and select Confidence Ellipsis to add the desired classifier. Select Color Data Points, assign colors, and click OK to apply changes.

Under Preliminaries, select Set Options for the Shape Graph located at the bottom of the screen. Choose Wireframe Graphs and modify the colors of the target shape, starting shape, and numbers. Select variation, then choose procrustes ANOVA.

Export the procrustes ANOVA results to the previously created spreadsheet file. In MorphoJ, Select the original dataset in Project Tree. Click Comparison, then Canonical Variate Analysis.

Select the classifier variables GENOTYPE TREATMENT and execute the function. To export the results, click the Results tab and right-click on the results page. Select Export to File and save the results.

In Project Tree, select Canonical Variate Analysis, then scores. Click File, choose Export Dataset, select the data type and GENOTYPE TREATMENT. Save the CVA scores as a TXT file.

To prepare the file for PAST software, open the saved CVA scores in a text editor. Replace the term id in the top-left corner with Label and save the edited file. Open the PAST software and click File, then open to select the edited CVA scores file.

In the import window. Select Names, data for row and column and Tab as the separator, then click Import. Under the Show tab, select Column attributes.

In the dropdown menu next to Type, assign Group to the first column containing classifier variables. Highlight the principal component or canonical variate data columns. Select Multivariate, then Tests, and choose Multivariate normality to run the normality test separately for principal component and canonical variate data.

Click the gray empty cell in the top-left corner above Type to select the entire dataset. Select Multivariate, then Tests, and choose multivariate analysis of variants to perform the analysis. Export the MANOVA results to the previously created spreadsheet file.

Images of the viscerocranium were acquired for each larva in each genotype and treatment group. Cartilages of the viscerocranium were labeled in a representative image, and landmarks were placed on each image to generate landmark-labeled data sets. Principal component 1 accounted for approximately 34%of total shape variation and principal component 2 accounted for 20%of total shape variation.

Each principal component showed specific variation in viscerocranial shape relative to the average shape of all viscerocrania. The principal component analysis plot showed overlapping means among the genotype and treatment groups with no distinct clustering. Ethanol-treated wild-type larvae exhibited a large 95%confidence ellipse of the mean due to the low sample size.

Canonical variate 1 represented a subtle shortening or lengthening of the ceratohyal joint in the magenta wireframe relative to the average black wireframe. Canonical variate 2 showed a shift medially only on one side of the viscerocranium at the joints between the Meckel's-palatoquadrate and the palatoquadrate-ceratohyal. Ethanol-treated wild-type larvae exhibited a large 95%confidence ellipse of the mean that overlapped all other groups.

Multivariate analysis of variants revealed a significant overall effect of genotype and treatment. Consistent sample mounting is the biggest challenge. Tilted larva can cause inaccurate measurements and alter results.

This protocol adapts to different anatomical structures, analyzes 3D data, generates linear measures, and helps determine experimental sample sizes. Future studies will investigate other genotype-ethanol sensitivities, expand sample sizes, and apply this versatile toolbox to various anatomical structures.