Files
Download Full Text (14.9 MB)
Abstract
The function of around 67% of predicted proteins from genes in Mycobacteriophage CheetoDust can not be confidently predicted using traditional techniques and can only be functionally labeled “hypothetical proteins”. However, a new approach using AlphaFold, an artificial intelligence tool to generate a structural prediction from a sequence, can take advantage of structurally conserved regions that were previously obfuscated to gain new insights and visualize data in new ways.
Since amino acid sequences are more conserved than its corresponding DNA sequence, amino acid sequences are used when predicting the function of the corresponding translated protein. Until recently, predicting structure from an amino acid sequence, known as the “protein-folding problem,” was unreliable, and thus, not widely used. Functional analysis of predicted proteins can now be enhanced with new tools like AlphaFold that attempt to solve the protein-folding problem and generate monomer and multimer structural predictions that can be used to further functional predictions.
For this project, AlphaFold was used to: 1) confirm the function of gene product 68 as DNA primase/helicase, 2) confirm a ‘wing’ structure to enable functional annotation as a winged-helix-turn-helix protein, and 3) predict functions for other proteins previously annotated as hypothetical protein. This poster will explore the potential for using artificial intelligence to improve functional annotation of protein sequences.
Publication Date
2023
Subject Major(s)
Bioinformatics
Keywords
phage, bioinformatics, protien, proteome, AlphaFold, FolkSeek, function, prediction, art, chimerax
Disciplines
Bioinformatics | Structural Biology
Faculty Advisor/Mentor
Allison Johnson
Rights
© The Author(s)