Computational anatomy and geometric shape analysis enables analysis of complex craniofacial phenotypes in zebrafish

Abstract

Due to the complexity of fish skulls, previous attempts to classify craniofacial phenotypes have relied on qualitative features or sparce 2D landmarks. In this work we aim to identify previously unknown 3D craniofacial phenotypes with a semiautomated pipeline in adult zebrafish mutants. We first estimate a synthetic 'normative' zebrafish template using MicroCT scans from a sample pool of wild-type animals using the Advanced Normalization Tools (ANTs). We apply a computational anatomy (CA) approach to quantify the phenotype of zebrafish with disruptions in bmp1a, a gene implicated in later skeletal development and whose human ortholog when disrupted is associated with Osteogenesis Imperfecta. Compared to controls, the bmp1a fish have larger otoliths, larger normalized centroid sizes, and exhibit shape differences concentrated around the operculum, anterior frontal, and posterior parietal bones. Moreover, bmp1a fish differ in the degree of asymmetry. Our CA approach offers a potential pipeline for high-throughput screening of complex fish craniofacial shape to discover novel phenotypes for which traditional landmarks are too sparce to detect. The current pipeline successfully identifies areas of variation in zebrafish mutants, which are an important model system for testing genome to phenome relationships in the study of development, evolution, and human diseases. This article has an associated First Person interview with the first author of the paper.

Keywords: Computational anatomy; Cranial morphology; Geometric morphometrics; Osteogenesis imperfecta.

Fig. 1.

Pipeline validation using 23 traditional landmarks. Boxplots are shown for (A) Euclidean distance between Gold Standard landmark locations and ALPACA transferred points (orange), and distance between the two manual landmark placements (grey), where color indicates method used for all panels, and (B) normalized centroid size of gold standard and ALPACA transferred pseudo-landmarks. Midline of boxplots show median value, with hinges corresponding to first and third quartiles, and whiskers extending to largest and smallest value no further than 1.5 times the interquartile range. Also shown (C) are the first two principal components of shape space from the combined GPA analysis of gold standard and ALPACA transferred 23 landmarks. The percent of variance for each PC is show in parenthesis of each axis. Individual fish are depicted as different numbers, fish 1–12 are crispant fish, fish 13–26 are wild-type fish, and fish 27 is the atlas.

Fig. 2.

Heat map of (A) symmetric and (B) asymmetric components of shape variation. Lateral and anterior views are shown for each group (wild type and bmp1a) within both components of shape variation. Colors show variation in shape from the symmetric atlas, with deeper colors representing greater variation from the atlas.

Fig. 3.

First two principal components of symmetry analysis. PC plots show separation of groups (represented by color) along the first and second PCs (A,B). Heat maps of the same PCs represent where shape variation occurs across each axis (C,D). Columns represent symmetric (A,C) and asymmetric (B,D) components of shape variation. The central image in C and D represents mean shape of each component. Color in C and D represents the Procrustes distance between the average shape and the shape occupying the ends of each PC axis. Deeper colors represent larger differences, and the specific colors refer to differences in direction relative to the average image.

Fig. 4.

Gold standard of 23 manual landmarks. Landmark placement represents the average location of two independent landmark placements by the same author. Right lateral (A), dorsal (B), ventral (C), and left lateral (D) views of the atlas mesh are shown. Landmark definitions can be found in Table 3.

Fig. 5.

Pipeline for atlas building, pseudo-landmark generation, and transferring pseudo-landmarks to individual fish. Blue text notes the software used between each step. (1) Starting with µCT scans of wild-type fish, ANTs uses a series of rigid, affine, and deformable registrations to create an average image, or (2) Atlas. The PseudoLMGenerator tool in SlicerMorph was used to (3) place 372 pseudo-landmarks on the atlas. The ALPACA tool in SlicerMorph was used to (4) transfer points from the atlas to wild-type and bmp1a fish for comparisons between groups.