TY - GEN
T1 - TSSNet – A Deep Neural Network Model for Predicting Prokaryotic Transcription Start Sites
AU - Ni, Chung En
AU - Doan, Duy Phuong
AU - Chiu, Yen Jung
AU - Huang, Yen Hua
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Several computational methods have been developed to perform promoter classification in prokaryotic genomes. However, those methods were mostly designed to classify promoters only under the condition that the TSS must be located at a predefined anchor position in each input sequence. Hence, this study aims at developing a new method that can classify promoters without assuming a fixed location of TSS. We draw an analogy between TSS identification and object detection, which is a well-known task in image analysis. Thus, TSSNet, a deep neural network model, is developed in this study and it can scan for the TSSs by taking windowed regions in a prokaryotic genome. The benchmark reveals that TSSNet has a recall rate higher than 80% in TSS identification when analyzing the genomic sequences. Our results suggest that an object detection model can be applied to the analysis of genomic sequences for finding the regulatory elements and key points that are important in functional genomics.
AB - Several computational methods have been developed to perform promoter classification in prokaryotic genomes. However, those methods were mostly designed to classify promoters only under the condition that the TSS must be located at a predefined anchor position in each input sequence. Hence, this study aims at developing a new method that can classify promoters without assuming a fixed location of TSS. We draw an analogy between TSS identification and object detection, which is a well-known task in image analysis. Thus, TSSNet, a deep neural network model, is developed in this study and it can scan for the TSSs by taking windowed regions in a prokaryotic genome. The benchmark reveals that TSSNet has a recall rate higher than 80% in TSS identification when analyzing the genomic sequences. Our results suggest that an object detection model can be applied to the analysis of genomic sequences for finding the regulatory elements and key points that are important in functional genomics.
KW - deep neural network
KW - object detection
KW - promoter classification
KW - transcription start sites
UR - http://www.scopus.com/inward/record.url?scp=85145591393&partnerID=8YFLogxK
U2 - 10.1109/BIBE55377.2022.00054
DO - 10.1109/BIBE55377.2022.00054
M3 - Conference contribution
AN - SCOPUS:85145591393
T3 - Proceedings - IEEE 22nd International Conference on Bioinformatics and Bioengineering, BIBE 2022
SP - 221
EP - 224
BT - Proceedings - IEEE 22nd International Conference on Bioinformatics and Bioengineering, BIBE 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 22nd IEEE International Conference on Bioinformatics and Bioengineering, BIBE 2022
Y2 - 7 November 2022 through 9 November 2022
ER -