Trait-Predicting Algorithms: Proposed Algorithm for Predicting Phenotypes from Genetic Data; Social and Ethical Attitudes Towards Trait Prediction via Genomic Samples

This portfolio presents a comprehensive study on trait-predicting algorithms, investigating both their construction and ethical implications. The STS Research Paper focuses on the ethical dilemmas associated with these algorithms, including privacy concerns and racial bias. A rigorous analysis of existing literature and legislation on genetic privacy policies, as well as a poll of UVA students' opinions, were employed to draw conclusions on how these algorithms should be implemented, and what regulations are necessary to ensure their safe use. The technical paper presents a conceptual algorithm for trait prediction, a task previously performed manually for a limited number of phenotypes. The algorithm leverages a neural network structure to identify complex phenotypes in genetic data. The paper discusses the construction and training of the algorithm in detail, covering aspects such as the ideal training set, data structuring, and the optimal system for training. Combined, these two papers create a cohesive solution for researchers interested in working with, or building trait-predicting algorithms in an ethical manner.