Using AI to Predict Protein Structural StabilityPublic
Downloadable Contentopen in viewer
Alternative splicing contributes significantly to proteome diversity in humans. Yet, many classified alternatively spliced isoforms lack physical evidence of their physiological existence in the human cell. This is likely because these isoform amino acid sequences create structurally unstable proteins that are immediately degraded. To allow for a better understanding of the human proteome’s diversity, a Positive-Unlabeled learning classification algorithm was implemented to accurately predict the existence of a protein with protein structural stability being the determining factor. This algorithm was given features consisting of protein data relevant to structural stability and was tested/trained on verified structurally stable proteins and proteins with unknown structural stability. To quantifiably demonstrate its predicting power, the algorithm was then used to predict the structural stability of proteins from the genes CFTR and TP53 that were physically confirmed as structurally stable and unstable. Improvements are necessary, but good results during testing/training were acquired and the algorithm effectively predicted the structural stability of the proteins from CFTR and TP53.
- This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.
- Date created
- Resource type
- Rights statement
- In Collection:
|Thumbnail||Title||Visibility||Embargo Release Date||Actions|
|MQP Presentation Poster_Final_0.pdf||Public||Download|
|_Final Draft_ MQP Report_0.pdf||Public||Download|
Permanent link to this page: https://digital.wpi.edu/show/dn39x448x