DNAImpute wins Rep. Lucy McBath’s 2022 Congressional App Challenge in Georgia’s Sixth District

When asked what inspired the creation of DNAImpute, the students said, ”First, we’ll discuss why we created a DNA imputation algorithm. The sequencing of genomes has helped researchers gain an accurate understanding of the biology of various organisms. However, due to the fragility of DNA, current DNA sequencing methods are not completely accurate and data is prone to be degraded, thus resulting in lost data. Due to these gaps in DNA data, researchers are unable to analyze genomes to their full extent. Some models and tools have been created for DNA imputation that are able to predict what DNA base pairs would be present in these blank spaces. Current methods are however, not entirely accurate and require supplementary data, apart from the raw genome data, to generate accurate predictions. This is why we attempted to create an imputation model using machine learning algorithms that works solely on genome data of the target organisms along with some reference organisms that are closely related to the target.

Although our algorithm worked extremely well (>85% accuracy), we wanted users to be able to visualize the imputations and how they came about. Thus, we created a web app in which users can play through each prediction while viewing pre- and post-sequence data and viewing Levenshtein distances (our metric for DNA sequence similarity). We accomplished this in our interactive web app and were also able to add a sign-in feature, so that users could store their data in their accounts and view reports, logs, and statistics for each imputation.”