INitroY-Deep: Computational Identification of Nitrotyrosine Sites to Supplement Carcinogenesis Studies Using Deep Learning
In biological systems, Nitration is a crucial post-translational modification which occurs on various amino acids. Nitration of Tyrosine is regarded as nitorsative stress biomarker resulting in the formation of peroxynitrite and other reactive and harmful nitrogen species. NitroTyrosine is closely r...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Published: |
Institute of Electrical and Electronics Engineers Inc.
2021
|
| Online Access: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85105890893&doi=10.1109%2fACCESS.2021.3080041&partnerID=40&md5=4b111f85a40c1da482d8af011e6b7f43 http://eprints.utp.edu.my/23887/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In biological systems, Nitration is a crucial post-translational modification which occurs on various amino acids. Nitration of Tyrosine is regarded as nitorsative stress biomarker resulting in the formation of peroxynitrite and other reactive and harmful nitrogen species. NitroTyrosine is closely related to Carcinogenesis, tumor growth progression and other major pathological conditions including systemic autoimmune diseases, inflammation, neurodegeneration and cardiovascular disorders. Additionally, the alteration in Nitrotyrosine profile occurs well before appearance of any symptoms of aforementioned diseases making nitrotyrosine a biomarker and potential target for early prognosis of aforementioned diseases. The wet lab identification of potential nitrotyrosine sites is laborious, time-taking and costly due to challenges of in vitro, ex vivo and in vivo identification processes. To supplement wet lab identification of nitrotyrosine, we proposed, implemented and evaluated a different approach to develop tyrosine nitration site predictors using pseudo amino acid compositions (PseAAC) and deep neural networks (DNNs). Proposed approach does not require any feature extraction and uses DNNs for learning a feature representation of peptide sequences and classification thereof. Validation of proposed approach is done using well-known model evaluation measures. Among different deep neural networks, convolutional neural network-based predictor achieved best scores on independent dataset with accuracy of 87.2, matthew's correlation coefficient score of 0.74 and AuC score of 0.91 which outperforms the previous reported scores of Nitrotyrosine predictors. © 2013 IEEE. |
|---|