단백질 기능 예측 문제에서 시퀀스 패턴 추출을 위한 작은 CNN-RNN 접목 모델 연구
Small CNN-RNN Engraft Model Study for Sequence Pattern Extraction in Protein Function Prediction Problems
2
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
본 논문에서는 2020년 기준 단백질 서열을 이용한 기능과 구조 예측 분야에서 가장 많이 사용되고있는 딥러닝 모델인 CNN과 LSTM/GRU 모델을 동일한 조건 하에 비교 평가한 연구를 토대로 새로운효소 기능 예측 모델인 PSCREM을 설계하였다. CNN 합성곱 시 누락되는 세부 패턴을 보존하기 위하여서열 진화정보를 이용하였으며 중첩 RNN을 통해 기능적으로 중요한 의미를 가지는 아미노산 간의관계 정보를 추출하고 특징 맵 제작에 참조하였다. 사용된 RNN 계열의 알고리즘은 LSTM과 GRU로보통 stacked RNN 기법으로 100 units 이상 2~3회 쌓는 것이 일반적이나 본 논문에서는 10, 20 unit으로구성한 뒤 중첩시켜서 특징 맵 제작에 사용하였다. 모델에 들어가는 데이터는 단백질 서열 데이터로PSSM profile로 가공한 뒤 사용되었다. 실험 결과 효소 번호 첫 번째 자리를 예측하는 문제에 대해86.4%의 정확도를 나타냄을 입증하였고, 효소 번호 3번째 자리까지 예측 정확도 84.4%의 성능을 내는것을 확인하였다. PSCREM은 Overlapped RNN을 통해 단백질 기능에 관련된 고유 패턴을 더 잘 파악하며Overlapped RNN은 단백질 기능 및 구조 예측 추출 분야에 새로운 방법론으로서 제안된다.
Abstract
In this paper, we designed a new enzyme function prediction model PSCREM based on a study that compared and evaluated CNN and LSTM/GRU models, which are the most widely used deep learning models in the field of predicting functions and structures using protein sequences in 2020, under the same conditions. Sequence evolution information was used to preserve detailed patterns which would miss in CNN convolution, and the relationship information between amino acids with functional significance was extracted through overlapping RNNs. It was referenced to feature map production. The RNN family of algorithms used in small CNN-RNN models are LSTM algorithms and GRU algorithms, which are usually stacked two to three times over 100 units, but in this paper, small RNNs consisting of 10 and 20 units are overlapped. The model used the PSSM profile, which is transformed from protein sequence data. The experiment proved 86.4% the performance for the problem of predicting the main classes of enzyme number, and it was confirmed that the performance was 84.4% accurate up to the sub-sub classes of enzyme number. Thus, PSCREM better identifies unique patterns related to protein function through overlapped RNN, and Overlapped RNN is proposed as a novel methodology for protein function and structure prediction extraction.
참고문헌 (21)
- Y. Liang, "Prediction of Protein Structural Classes for Low-Similarity Sequences Based on Consensus Sequence and Segmented PSSM", Computational and Mathematical Methods in Medicine, 2015, pp.9~ (2015)[DOI]
- J. Wang, "POSSUM: a bioinformatics toolkit for generating numerical sequence feature descriptors based on PSSM profiles", Bioinformatics, 33(17), pp.2756~2758 (2017)[DOI]
- Mousavian Z, "Drug-target interaction prediction from PSSM based evolutionary information", Journal of pharmacological and toxicological methods, 78, pp.42~51 (2016)[DOI]
- N. Q. K. Le, "SNARE-CNN: a 2D convolutional neural network architecture to identify SNARE proteins from high-throughput sequencing data", PeerJ. Computer science, 5, pp.e177~ (2019)[DOI]
- Y. Guo, "EPTool: A New Enhancing PSSM Tool for Protein Secondary Structure Prediction", Journal of computational biology : a journal of computational molecular cell biology, 28, pp.362~364 (2021)[DOI]
- Liu Y, "SNB-PSSM: A spatial neighbor-based PSSM used for protein-RNA binding site prediction", J Mol Recognit, 34, pp.e2887~ (2021)[DOI]
- A. Dalkiran, "ECPred:a tool for the prediction of the enzymatic functions of protein sequences based on the EC nomenclature", BMC bioinformatics, 19, pp.334~ (2018)[DOI]
- A. Amidi, "EnzyNet: enzyme classification using 3D convolutional neural networks on spatial representation", PeerJ, 6, pp.e4750~ (2018)[DOI]
- X. Xiao, "MF-EFP: Predicting Multi-Functional Enzymes Function Using Improved Hybrid Multi-Label Classifier", IEEE Access, 8, pp.50276~50284 (2020)[DOI]
- N. Strodthoff, "UDSMProt: universal deep sequence models for protein classification", Bioinformatics, 36(8), pp.2401~2409 (2020)[DOI]
- 이정민, "단백질 기능 예측 모델의 주요 딥러닝 모델 비교 실험", 정보처리학회 논문지, 11(6), pp.245~254 (2022), 한국정보처리학회
- Suzuki H, "How Enzymes Work: From Structure to Function", pp.117~140 (2015), CRC Press
- D. M. Debra, "Enzyme function discovery", Structure, 16(11), pp.1599~1600 (2008)[DOI]
- Saigo, Hiroto, "Reaction graph kernels predict EC numbers of unknown enzymatic reactions in plant secondary metabolism", BMC Bioinformatics, 11(Suppl 1), pp.S31~ (2010)[DOI]
- A. G. McDonald, "Enzyme nomenclature and classification: the state of the art", FEBS J (2021)[DOI]
- A. A. Schäffer 1, "Improving the accuracy of PSI-BLAST protein database searches with composition-based statistics and other refinements", Nucleic Acids Res, 29(14), pp.2994~3005 (2001)[DOI]
- S. Kim, "Basic for Protein Structure Prediction: BLAST and Profile", Biophysical Society Newsletter, 11(1) (2005)
- Y. Kim, "Convolutional Neural Networks for Sentence Classification", Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp.1746~1751 (2014)
- J. Y. Ryu, "Deep learning enables high-quality and high-throughput prediction of enzyme commission numbers", Proceedings of the National Academy of Sciences of the United States of America, 116(28), pp.13996~14001 (2019)[DOI]
- Gao, Ruibo, "Prediction of Enzyme Function Based on Three Parallel Deep CNN and Amino Acid Mutation", International journal of molecular sciences, 20(11), pp.2845~ (2019)[DOI]
- A. L. Rio, "Effect of sequence padding on the performance of deep learning models in archaeal protein functional prediction", Scientific Reports, 10(1), pp.14634~ (2020)[DOI]