Hosted by: Jose Dinneny
The shapes around us contain information. Morphometrics, methods that quantify shape, can reveal patterns reflecting underlying biological phenomena. In this seminar, I will begin with a primer to traditional morphometric methods using violins as an example. The outlines of >9,000 instruments sold at auction over 400 years of history are decomposed using Elliptical Fourier Descriptors, a Fourier-based method that treats outlines as waves. Predictive modeling reveals those violin makers that imitate others, and clustering methods recapitulate the genetic lineages of luthiers. As useful as morphometric methods are for shapes (like violins and leaves) they are not appropriate for the branching structures of plants—for example trees, inflorescences, and roots. I will end with a discussion of a promising topological technique, persistent homology, that unlike traditional morphometric approaches, can capture the branching architectures of plants. I will describe the application of persistent homology to leaf shape, leaflet serrations, and root architecture in the same plants that reveals a shared genetic basis for these diverse features in tomato introgression lines. I will end with a preview of applying persistent homology to predict plant family and location independently of each other in >170,000 leaves, and quantifying the complex, 3D branching architectures of roots and inflorescences using X-ray Computed Tomography.