Neural Network Analysis of Biomolecular Representations in Vertical Translation of Protein Structure in Experts and Students
Several studies have reported that without intentional and strategic instruction across molecular life science curricula, development of visual literacy skills occurs independently from instruction. Research with validated instruments to assess the organization of student and expert neural networks of visual literacy skills remains to be accomplished. Here we report the development and results of a tool that measures the vertical translation of protein structure in undergraduates. Expert and student participants' neural networks were generated from a survey where they ranked the relatedness of biochemical representations related to protein structure. Using the Pathfinder program, neural network images were generated along with the coherency, neighborhood similarity (NS), and path length correlation (PLC) values for each expert and student participant. In order to determine and compare the validity of student results, these data were analyzed against an established expert referent network created from the expert participants. The student participants were enrolled in chemistry and biochemistry courses from a large, R-1 university and a small, PUI. Our preliminary results compare the neural network organization of vertical translation of protein structure between third and fourth year students at these institutions. Analysis of the coherency, PLC, and NS variables suggest strategic curriculum modifications that could aid in the development of expert-like visual literacy skills for students in molecular life science courses, by enabling students to better organize their vertical translation of protein structure’ neural networks.