Autoencoders for Signature Extraction: Systematically evaluating Pre- and Post-Processing
Term
4. term
Education
Publication year
2024
Submitted on
2024-06-10
Pages
20
Abstract
Cancer is fundamentally a genetic disorder caused by various differ- ent factors. Each mutation within the cancer genome leaves a unique and identifiable signature in the DNA sequence. These signatures are the focus of this paper, and how they can be used to identify the cause. In this paper, we aim to enhance the performance and accuracy of extracted signatures using autoencoders. To enhance the perfor- mance we explore and analyze different pre- and post-processing methods and compare the results to a baseline. To determine the performance of the different methods, by comparing the extracted signatures to known signatures from COSMIC and Signal. The com- parison uses cosine similarity as the metric, and then later plotted for visualization of the results. The findings from the study showed that the added steps in the pipeline had a good effect and increased the performance and accuracy of the extracted signatures.
Documents