@techreport{oai:ipsj.ixsq.nii.ac.jp:00234923,
 author = {Md., Harun-Or-Roshid and Hiroyuki, Kurata and Md., Harun-Or-Roshid and Hiroyuki, Kurata},
 issue = {36},
 month = {Jun},
 note = {Dihydrouridine (DHU, D) is a prevalent post-transcriptional modification in tRNA, mRNA, and snoRNA, linked to disease and biological processes in eukaryotes. Identifying D sites is crucial but experimental methods are costly and slow. To address this, we developed a computational tool that enhances prediction performance by integrating 66 baseline models through ensemble learning, named Stack-DHUpred. These models combine six machine-learning classifiers with eleven feature encoding methods. The best-performing combinations were used to construct the final stacked model. In tests, Stack-DHUpred surpassed existing predictors on the independent dataset, proving the efficacy of our stacking approach in accelerating the discovery and understanding of D modifications in post-transcriptional regulation., Dihydrouridine (DHU, D) is a prevalent post-transcriptional modification in tRNA, mRNA, and snoRNA, linked to disease and biological processes in eukaryotes. Identifying D sites is crucial but experimental methods are costly and slow. To address this, we developed a computational tool that enhances prediction performance by integrating 66 baseline models through ensemble learning, named Stack-DHUpred. These models combine six machine-learning classifiers with eleven feature encoding methods. The best-performing combinations were used to construct the final stacked model. In tests, Stack-DHUpred surpassed existing predictors on the independent dataset, proving the efficacy of our stacking approach in accelerating the discovery and understanding of D modifications in post-transcriptional regulation.},
 title = {Identification of Dihydrouridine RNA modification sites through stacking strategy},
 year = {2024}
}