Interpretable Deep Learning for Studying Transcriptional and Post-Transcriptional Regulations

Simple item page
dc.contributor.advisorZhang, Jianqiu
dc.contributor.authorZhang, Tinghe
dc.contributor.committeeMemberFlores, Mario
dc.contributor.committeeMemberJin, Yufang
dc.contributor.committeeMemberChen, Yidong
dc.date.accessioned2024-03-08T17:40:54Z
dc.date.available2024-03-08T17:40:54Z
dc.date.issued2022
dc.descriptionThis item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.
dc.description.abstractTranscription regulation and post-transcription regulation are critical biological processes for organisms' development, complexity, and homeostasis. Understanding the mechanisms of these processes will be helpful for biologists to reveal the secret of life. Traditionally, biological discoveries are achieved mainly by experiments. However, the experiments are costly and time-consuming. Developing computation tools that elucidate biological functions from data can accelerate biological discovery. In this study, we focused on three topics to investigate functional predictions of three different phases of Transcription and post-transcription regulation by interpretable deep learning methods. We first considered the prediction of enhancers, which are cis-acting DNA regulatory regions that play a key role in increasing the transcription of specific genes via interaction with transcription factors. We designed a CNN-based residual neural network to identify enhancers and their strength. A 4% accuracy improvement in independent tests shows that the proposed model can effectively predict the enhancer's strength. Then, we investigated the prediction of YTHDF2-mediated mRNA degradation based on mRNA sequences and proposed m6ABERT, a transformer-based model. Our models reported at least 2.5% improvement in accuracy than other models. Besides, we discovered the potential RNA binding proteins that affect the degradation by interpreting m6ABERT.For gene expression, we proposed an interpretable gene expression-based deep learning model, T-GEM, for phenotype prediction and gene regulatory network discovery. We showed the competitive performance with existing models and the advantage of the model's interpretability. We also revealed the learning mechanism of T-GEM and devised a method to extract the regulatory network from T-GEM.
dc.description.departmentElectrical and Computer Engineering
dc.format.extent92 pages
dc.format.mimetypeapplication/pdf
dc.identifier.isbn9798358492011
dc.identifier.urihttps://hdl.handle.net/20.500.12588/6204
dc.languageen
dc.subjectDeep learning
dc.subjectTranscription regulation
dc.subjectmRNA degradation
dc.subjectGene expression
dc.subject.classificationElectrical engineering
dc.titleInterpretable Deep Learning for Studying Transcriptional and Post-Transcriptional Regulations
dc.typeThesis
dc.type.dcmiText
dcterms.accessRightspq_closed
thesis.degree.departmentElectrical and Computer Engineering
thesis.degree.grantorUniversity of Texas at San Antonio
thesis.degree.levelDoctoral
thesis.degree.nameDoctor of Philosophy

Files

Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
Zhang_utsa_1283D_13797.pdf
Size:
2.97 MB
Format:
Adobe Portable Document Format
Download

Collections

Electronic Theses and Dissertations - UTSA Access Only