PLACE: 1104 Gilman

SPEAKER:
Dean Adams

TITLE:
The Statistical Analysis of Biological Shape

ABSTRACT:

Biologists have long thought that form and function are closely linked, and this form-function relationship is believed to hold true at many levels of biological organization; from anatomical comparisons (functional anatomy and biomechanics) to molecular level comparisons (protein structure and function). Crucial to any form-function analysis however, is a proper mathematical description of biological form (shape).  Such a description of biological shape falls under the purview of morphometrics.  In this seminar I will review some
of the commonly used methods for quantifying biological shape.  I will describe the classical multivariate morphometric techniques, demonstrate their use, and explain their major shortcomings.  I will then describe a relatively new set of statistical tools (geometric morphometric methods) that circumvent these same difficulties, and present several examples of their use in anatomical studies. Finally, I will describe how these geometric methods may be applied to the analysis of protein structures.  Because it is believed that identifying structurally similar proteins is the first step in identifying functionally similar proteins, it is important to use methods that are mathematically consistent and efficient in their representation of biological shape.  I will show that for a representative data set of globin proteins, the geometric methods borrowed from comparative anatomy perform markedly better than the currently accepted bioinformatic methods for comparing protein structures (the DALI Dictionary).
 
 

COFFEE: 3:45 p.m., 104 Snedecor Hall