Study progress of protein function prediction based on sequence information and machine learning

doi:10.16178/j.issn.0528-9017.20250739

Abstract

Abstract:

With the rapid advancement of computational power and the expansion of biological data，protein sequencing technology has seen significant progress，leading to a sharp increase in the number of proteins within databases. Due to the complexity and diversity of protein data，functional prediction has become a highly challenging issue and has garnered widespread attention. Concurrently，with the rapid development of artificial intelligence，methods such as machine learning have been progressively applied to protein function prediction. In recent years，researchers domestically and internationally have been continuously exploring this field，achieving substantial research outcomes. This paper summarized the functional prediction methods of protein sequence information based on bioinformatics. Furthermore，we analyzed and summarized the specific algorithms and recent advances within these approaches. Finally，we discussed existing challenges in protein function prediction and provided an outlook on future research directions in this domain.

Key words: bioinformatics, machine learning, protein, function prediction

CLC Number:

WU Jianxiong, BAO Yufeng. Study progress of protein function prediction based on sequence information and machine learning[J]. Journal of Zhejiang Agricultural Sciences, 2026, 67(2): 552-556.

Figures/Tables 4

References 23

[1]	ASHBURNER M， BALL C A， BLAKE J A，et al. Gene ontology：tool for the unification of biology：the gene ontology consortium［J］. Nature Genetics，2000，25（1）：25-29.
[2]	BOADU F， CAO H Y， CHENG J L. Combining protein sequences and structures with transformers and equivariant graph neural networks to predict protein function［J］. Bioinformatics，2023，39（）：i318-i325.
[3]	TIWARI A K， SRIVASTAVA R. A survey of computational intelligence techniques in protein function prediction［J］. International Journal of Proteomics，2014，2014：845479.
[4]	ZHOU N H， JIANG Y X， BERGQUIST T R，et al. The CAFA challenge reports improved protein function prediction and new functional annotations for hundreds of genes through experimental screens［J］. Genome Biology，2019，20：244.
[5]	DAVID J LIPMAN W R P. Rapid and sensitive protein similarity searches［J］. Science，1985，227（4693）：1435-1441.
[6]	ALTSCHUL S F， GISH W， MILLER W，et al. Basic local alignment search tool［J］. Journal of Molecular Biology，1990，215（3）：403-410.
[7]	ALTSCHUL S F， MADDEN T L， SCHÄFFER A A，et al. Gapped BLAST and PSI-BLAST：a new generation of protein database search programs［J］. Nucleic Acids Research，1997，25（17）：3389-3402.
[8]	REIJNDERS M J M F. Wei2GO：weighted sequence similarity-based protein function prediction［J］. PeerJ，2022，10：e12931.
[9]	LAN L， DJURIC N， GUO Y H，et al. MS-kNN：protein function prediction by integrating multiple data sources［J］. BMC Bioinformatics，2013，14（）：S8.
[10]	GLIGORIJEVIĆ V， JANJIĆ V， PRŽULJ N. Integration of molecular network data reconstructs gene ontology［J］. Bioinformatics，2014，30（17）：i594-i600.
[11]	蔡从中，韩连漪，王万录，等. 支持向量机程序SVMProt预测SARS病毒蛋白质的功能［J］. 重庆大学学报（自然科学版），2003，26（9）：148-150.
	CAI C Z， HAN L Y， WANG W L，et al. Prediction of the function of SARS proteins by using a support vector machine program SVMProt［J］. Journal of Chongqing University （Natural Science Edition），2003，26（9）：148-150.
[12]	RANJAN A， TIWARI A， DEEPAK A. A sub-sequence based approach to protein function prediction via multi-attention based multi-aspect network［J］. IEEE/ACM Transactions on Computational Biology and Bioinformatics，2023，20（1）：94-105.
[13]	ZHENG L Y， SHI S Y， LU M K，et al. AnnoPRO：a strategy for protein function annotation based on multi-scale protein representation and a hybrid deep learning of dual-path encoding［J］. Genome Biology，2024，25（1）：41.
[14]	WAN C， JONES D T. Protein function prediction is improved by creating synthetic feature samples with generative adversarial networks［J］. Nature Machine Intelligence，2020，2（9）：540-550.
[15]	LAI B Q， XU J B. Accurate protein function prediction via graph attention networks with predicted structure information［J］. Briefings in Bioinformatics，2022，23（1）：bbab502.
[16]	秦琪琪，丁学明，王金雷. 利用序列和组合图卷积网络预测蛋白质功能［J］. 小型微型计算机系统，2023，44（12）：2692-2699.
	QIN Q Q， DING X M， WANG J L. Protein function prediction using sequences and combined graph convolutional networks［J］. Journal of Chinese Computer Systems，2023，44（12）：2692-2699.
[17]	YANG K K， FUSI N， LU A X. Convolutions are competitive with transformers for protein sequence pretraining［J］. Cell Systems，2024，15（3）：286-294.e2.
[18]	BOADU F， CHENG J L. Improving protein function prediction by learning and integrating representations of protein sequences and function labels［J］. Bioinformatics Advances，2024，4（1）：vbae120.
[19]	ZHAPA-CAMACHO F， TANG Z W， KULMANOV M，et al. Predicting protein functions using positive-unlabeled ranking with ontology-based priors［J］. Bioinformatics，2024，40（）：i401-i409.
[20]	YAN H Y， WANG S J， LIU H C，et al. GORetriever：reranking protein-description-based GO candidates by literature-driven deep information retrieval for protein function annotation［J］. Bioinformatics，2024，40（）：ii53-ii61.
[21]	WANG R H， JIANG Y， JIN J R，et al. DeepBIO：an automated and interpretable deep-learning platform for high-throughput biological sequence prediction，functional annotation and visualization analysis［J］. Nucleic Acids Research，2023，51（7）：3017-3029.
[22]	POLITANO G， BENSO A， REHMAN H U，et al. PRONTO-TK：a user-friendly protein neural network tool-kit for accessible protein function prediction［J］. NAR Genomics and Bioinformatics，2024，6（3）：lqae112.
[23]	WANG W K， SHUAI Y Y， YANG Q R，et al. A comprehensive computational benchmark for evaluating deep learning-based protein function prediction approaches［J］. Briefings in Bioinformatics，2024，25（2）：bbae050.

发表年份	方法	模型评估标准
发表年份	方法	BP	CC	MF
2020	DeepGOPlus	F_max=0.390	F_max=0.614	F_max=0.557
2021	NCL+mask BLAST	F₁=0.378	F₁=0.475	F₁=0.496
2022	Wei2GO	F_max=0.520	F_max=0.580	F_max=0.390

发表年份	方法	模型评估标准
发表年份	方法	BP	CC	MF
2020	DeepGOPlus	F_max=0.390	F_max=0.614	F_max=0.557
2021	NCL+mask BLAST	F₁=0.378	F₁=0.475	F₁=0.496
2022	Wei2GO	F_max=0.520	F_max=0.580	F_max=0.390

发表年份	方法	模型评估标准
发表年份	方法	BP	CC	MF
2018	GOLabeler	F_max=0.372 AUPR=0.236	F_max=0.586 AUPR=0.697	F_max=0.691 AUPR=0.549
2020	DeepAdd	F_max=0.345 AUC=0.896	F_max=0.547 AUC=0.958	F_max=0.516 AUC=0.912
2020	FFPred-GAN	F_max=0.567	F_max=0.755	F_max=0.750
2021	Global-ProtEnc	F_max=0.523	F_max=0.636	F_max=0.515
2022	FUTUSA	F₁=0.532	—	—
2022	GAT-GO	F_max=0.501 AUPR=0.381	F_max=0.542 AUPR=0.479	F_max=0.637 AUPR=0.660
2022	SPROF-GO	F_max=0.335 AUPR=0.247	F_max=0.725 AUPR=0.765	F_max=0.647 AUPR=0.622
2023	ProteInfer	F_max=0.647 AUPR=0.622	F_max=0.335 AUPR=0.247	F_max=0.647 AUPR=0.622
2023	PFP-SCGCN	F_max=0.706	F_max=0.796	F_max=0.787
2024	AnnoPRO	F_max=0.609 AUPR=0.574	F_max=0.746 AUPR=0.749	F_max=0.763 AUPR=0.755
2024	TransFew	F_max=0.449 AUPR=0.244	F_max=0.726 AUPR=0.455	F_max=0.666 AUPR=0.363
2024	PU-GO	F_max=0.556	F_max=0.734	F_max=0.569
2024	GORetriever	F_max=0.545	F_max=0.653	F_max=0.659

发表年份	方法	模型评估标准
发表年份	方法	BP	CC	MF
2018	GOLabeler	F_max=0.372 AUPR=0.236	F_max=0.586 AUPR=0.697	F_max=0.691 AUPR=0.549
2020	DeepAdd	F_max=0.345 AUC=0.896	F_max=0.547 AUC=0.958	F_max=0.516 AUC=0.912
2020	FFPred-GAN	F_max=0.567	F_max=0.755	F_max=0.750
2021	Global-ProtEnc	F_max=0.523	F_max=0.636	F_max=0.515
2022	FUTUSA	F₁=0.532	—	—
2022	GAT-GO	F_max=0.501 AUPR=0.381	F_max=0.542 AUPR=0.479	F_max=0.637 AUPR=0.660
2022	SPROF-GO	F_max=0.335 AUPR=0.247	F_max=0.725 AUPR=0.765	F_max=0.647 AUPR=0.622
2023	ProteInfer	F_max=0.647 AUPR=0.622	F_max=0.335 AUPR=0.247	F_max=0.647 AUPR=0.622
2023	PFP-SCGCN	F_max=0.706	F_max=0.796	F_max=0.787
2024	AnnoPRO	F_max=0.609 AUPR=0.574	F_max=0.746 AUPR=0.749	F_max=0.763 AUPR=0.755
2024	TransFew	F_max=0.449 AUPR=0.244	F_max=0.726 AUPR=0.455	F_max=0.666 AUPR=0.363
2024	PU-GO	F_max=0.556	F_max=0.734	F_max=0.569
2024	GORetriever	F_max=0.545	F_max=0.653	F_max=0.659