Transformer and gated recurrent unit-based de novo sequencing algorithm for phosphopeptides

doi:10.11772/j.issn.1001-9081.2025010060

Journal of Computer Applications ›› 2026, Vol. 46 ›› Issue (1): 297-304.DOI: 10.11772/j.issn.1001-9081.2025010060

• Frontier and comprehensive applications • Previous Articles Next Articles

Transformer and gated recurrent unit-based de novo sequencing algorithm for phosphopeptides

Lijin YAO¹, Di ZHANG¹, Piyu ZHOU², Zhijian QU¹, Haipeng WANG¹()

^1.School of Computer Science and Technology，Shandong University of Technology，Zibo Shandong 255049，China
^2.Academy of Mathematics and Systems Science，Chinese Academy of Sciences，Beijing 100190，China

Received:2025-01-15 Revised:2025-03-25 Accepted:2025-03-26 Online:2026-01-10 Published:2026-01-10
Contact: Haipeng WANG
About author:YAO Lijin， born in 1999， M. S. candidate. His research interests include deep learning， bioinformatics.
ZHANG Di， born in 1997， M. S. candidate. His research interests include deep learning， bioinformatics.
ZHOU Piyu， born in 1995， Ph. D. candidate. His research interests include machine learning， bioinformatics.
QU Zhijian， born in 1980， Ph. D.， associate professor. His research interests include optimization algorithms， machine learning.
Supported by:
National Key Research and Development Program of China(2022YFA1304603)

基于Transformer和门控循环单元的磷酸化肽从头测序算法

姚理进¹, 张迪¹, 周丕宇², 曲志坚¹, 王海鹏¹()

^1.山东理工大学计算机科学与技术学院，山东淄博 255049
^2.中国科学院数学与系统科学研究院，北京 100190

通讯作者: 王海鹏
作者简介:姚理进（1999—），男，山东聊城人，硕士研究生， CCF学生会员，主要研究方向：深度学习、生物信息学
张迪（1997—），男，山东临沂人，硕士研究生，主要研究方向：深度学习、生物信息学
周丕宇（1995—），男，山东淄博人，博士研究生，主要研究方向：机器学习、生物信息学
曲志坚（1980—），男，山东青岛人，副教授，博士，主要研究方向：优化算法、机器学习
基金资助:
国家重点研发计划项目(2022YFA1304603)

Abstract

Abstract:

Peptide sequencing using tandem mass spectrometry for proteolytically digested peptides （referred to as peptide identification） is a foundational technology in proteomics research. Current de novo peptide sequencing algorithms face challenges in identifying phosphopeptides accurately， which are of significant biological importance. The primary reason is the complex fragmentation patterns induced by phosphorylation， the frequent occurrence of neutral loss peaks， and the low abundance of phosphopeptides' mass spectrum in conventional mass spectrometric data. To address these issues， a Transformer and Gated Recurrent Unit （GRU）-based de novo sequencing algorithm for phosphopeptides was proposed， namely TGNovo. A spectrum graph was introduced in TGNovo to model the mass differences between peaks explicitly， guiding the Transformer encoder to capture spectral features. The Transformer module and the GRU module jointly model the association between spectral and amino acid sequence features and the dependencies among spectral peaks and amino acids， respectively， working in concert to achieve peptide reconstruction. Compared to the fully Transformer-based de novo sequencing algorithm Casanovo， TGNovo fully utilizes prior spectral information through the spectrum graph and GRU module， enhancing the model's ability to model spectrum graph. In evaluations of phosphopeptide fragments across species， TGNovo outperforms Casanovo with average improvements of 16.5 percentage points in peptide-level recall and 37.1 percentage points in amino acid-level recall. Additionally， experimental results on an immune peptide dataset show that TGNovo-identified high-confidence antigenic peptides cover 86% of the database search results.

Key words: de novo sequencing, Transformer, Gated Recurrent Unit (GRU), spectrum graph, phosphopeptide

摘要：

利用串联质谱对蛋白质酶切产生的肽段进行测序（称为肽鉴定）是蛋白质组学研究的支撑技术。现有肽段从头测序算法在鉴定具有重要生物学意义的磷酸化修饰肽段时准确度受限，主要原因是磷酸化修饰导致碎裂模式更复杂，易产生中性丢失峰，且磷酸化肽的质谱图在常规质谱数据中丰度较低。因此，提出基于Transformer和门控循环单元（GRU）的从头测序算法TGNovo。TGNovo引入谱峰连接图，显式建模谱峰间的质量差关系，指导Transformer编码器捕捉谱图特征。Transformer模块与GRU模块分别建模谱图与氨基酸序列的关联以及谱峰间与氨基酸间的依赖关系，二者协同工作以实现肽段重建。相较于完全基于Transformer的从头测序算法Casanovo， TGNovo通过谱峰连接图和GRU模块充分利用谱图先验信息，增强了模型对谱图的建模能力。在跨物种磷酸化肽段评测中， TGNovo在肽水平和氨基酸水平的召回率上比Casanovo分别平均提升了16.5和37.1个百分点；此外，在免疫肽数据集上的实验结果表明， TGNovo鉴定的高可信抗原肽覆盖了数据库搜索结果的86%。

关键词: 从头测序, Transformer, 门控循环单元, 谱峰连接图, 磷酸化肽

CLC Number:

TP391.9

Lijin YAO, Di ZHANG, Piyu ZHOU, Zhijian QU, Haipeng WANG. Transformer and gated recurrent unit-based de novo sequencing algorithm for phosphopeptides[J]. Journal of Computer Applications, 2026, 46(1): 297-304.

姚理进, 张迪, 周丕宇, 曲志坚, 王海鹏. 基于Transformer和门控循环单元的磷酸化肽从头测序算法[J]. 《计算机应用》唯一官方网站, 2026, 46(1): 297-304.

Figures/Tables 14

References 32

[1]	AEBERSOLD R， MANN M. Mass-spectrometric exploration of proteome structure and function ［J］. Nature， 2016， 537（7620）： 347-355.
[2]	MA B， JOHNSON R. De novo sequencing and homology searching ［J］. Molecular and Cellular Proteomics， 2012， 11（2）： No.O111.014902.
[3]	ENG J K， SEARLE B C， CLAUSER K R， et al. A face in the crowd： recognizing peptides through database search ［J］. Molecular and Cellular Proteomics， 2011， 10（11）： No.R111.009522.
[4]	BARTELS C. Fast algorithm for peptide sequencing by mass spectroscopy ［J］. Biomedical and Environmental Mass Spectrometry， 1990， 19（6）： 363-368.
[5]	MA B， ZHANG K， HENDRIE C， et al. PEAKS： powerful software for peptide de novo sequencing by tandem mass spectrometry ［J］. Rapid Communications in Mass Spectrometry， 2003， 17（20）： 2337-2342.
[6]	FRANK A， PEVZNER P. PepNovo： de novo peptide sequencing via probabilistic network modeling ［J］. Analytical Chemistry， 2005， 77（4）： 964-973.
[7]	FISCHER B， ROTH V， ROOS F， et al. NovoHMM： a hidden Markov model for de novo peptide sequencing ［J］. Analytical Chemistry， 2005， 77（22）： 7265-7273.
[8]	MA B. Novor： real-time peptide de novo sequencing software ［J］. Journal of the American Society for Mass Spectrometry， 2015， 26（11）： 1885-1894.
[9]	TRAN N H， ZHANG X， XIN L， et al. De novo peptide sequencing by deep learning ［J］. Proceedings of the National Academy of Sciences of the United States of America， 2017， 114（31）： 8247-8252.
[10]	QIAO R， TRAN N H， XIN L， et al. Computationally instrument-resolution-independent de novo peptide sequencing for high-resolution devices ［J］. Nature Machine Intelligence， 2021， 3（5）： 420-425.
[11]	牟长宁，王海鹏，周丕宇，等.基于图卷积神经网络的串联质谱从头测序［J］.计算机应用， 2021， 41（9）： 2773-2779.
	MOU C N， WANG H P， ZHOU P Y， et al. De novo peptide sequencing by tandem mass spectrometry based on graph convolutional neural network ［J］. Journal of Computer Applications， 2021， 41（9）： 2773-2779.
[12]	WU R， ZHANG X， WANG R， et al. Denovo-GCN： de novo peptide sequencing by graph convolutional neural networks ［J］. Applied Sciences， 2023， 13（7）： No.4604.
[13]	YILMAZ M， FONDRIE W E， BITTREMIEUX W， et al. Sequence-to-sequence translation from mass spectra to peptides with a transformer model ［J］. Nature Communications， 2024， 15： No.6427.
[14]	MAO Z， ZHANG R， XIN L， et al. Mitigating the missing-fragmentation problem in de novo peptide sequencing with a two-stage graph-based deep learning model ［J］. Nature Machine Intelligence， 2023， 5（11）： 1250-1260.
[15]	MACEK B， MANN M， OLSEN J V. Global and site-specific quantitative phosphoproteomics： principles and application ［J］. Annual Review of Pharmacology and Toxicology， 2009， 49： 199-221.
[16]	ZONG Y， WANG Y， YANG Y， et al. DeepFLR facilitates false localization rate control in phosphoproteomics ［J］. Nature Communications， 2023， 14： No.2269.
[17]	MARX H， LEMEER S， SCHLIEP J E， et al. A large synthetic peptide and phosphopeptide reference library for mass spectrometry-based proteomics ［J］. Nature Biotechnology， 2013， 31（6）： 557-564.
[18]	FERRIES S， PERKINS S， BROWNRIDGE P J， et al. Evaluation of parameters for confident phosphorylation site localization using an orbitrap fusion tribrid mass spectrometer ［J］. Journal of Proteome Research， 2017， 16（9）： 3448-3459.
[19]	SHTEYNBERG D D， DEUTSCH E W， CAMPBELL D S， et al. PTMProphet： fast and accurate mass modification localization for the trans-proteomic pipeline ［J］. Journal of Proteome Research， 2019， 18（12）： 4262-4272.
[20]	TAUMER C， GRIESBAUM L， KOVACEVIC A， et al. Parallel reaction monitoring on a Q Exactive mass spectrometer increases reproducibility of phosphopeptide detection in bacterial phosphoproteomics measurements ［J］. Journal of Proteomics， 2018， 189： 60-66.
[21]	BEKKER-JENSEN D B， KELSTRUP C D， BATTH T S， et al. An optimized shotgun strategy for the rapid generation of comprehensive human proteomes ［J］. Cell Systems， 2017， 4（6）： 587-599.e4.
[22]	MEUL T， BERSCHNEIDER K， SCHMITT S， et al. Mitochondrial regulation of the 26S proteasome ［J］. Cell Reports， 2020， 32（8）： No.108059.
[23]	WU X N， CHU L， XI L， et al. Sucrose-Induced Receptor Kinase 1 is modulated by an interacting kinase with short extracellular domain ［J］. Molecular and Cellular Proteomics， 2019， 18（8）： 1556-1571.
[24]	OSMAN S， MOHAMMAD E， LIDSCHREIBER M， et al. The Cdk8 kinase module regulates interaction of the mediator complex with RNA polymerase II ［J］. Journal of Biological Chemistry， 2021， 296： No.100734.
[25]	BASSANI-STERNBERG M， BRÄUNLEIN E， KLAR R， et al. Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry ［J］. Nature Communications， 2016， 7： No.13404.
[26]	CHI H， LIU C， YANG H， et al. Comprehensive identification of peptides in tandem mass spectra using an efficient open search engine ［J］. Nature Biotechnology， 2018， 36（11）： 1059-1061.
[27]	SHAO G， CAO Y， CHEN Z， et al. How to use open-pFind in deep proteomics data analysis？ — a protocol for rigorous identification and quantitation of peptides and proteins from mass spectrometry data ［J］. Biophysics Reports， 2021， 7（3）： 207-226.
[28]	COX J， MANN M. MaxQuant enables high peptide identification rates， individualized p.p.b.-range mass accuracies and proteome-wide protein quantification ［J］. Nature Biotechnology， 2008， 26（12）： 1367-1372.
[29]	GASSAWAY B M， LI J， RAD R， et al. A multi-purpose， regenerable， proteome-scale， human phosphoserine resource for phosphoproteomics ［J］. Nature Methods， 2022， 19（11）： 1371-1375.
[30]	AN Z， ZHAI L， YING W， et al. PTMiner： localization and quality control of protein modifications detected in an open search and its application to comprehensive post-translational modification characterization in human proteome ［J］. Molecular and Cellular Proteomics， 2019， 18（2）： 391-405.
[31]	REYNISSON B， ALVAREZ B， PAUL S， et al. NetMHCpan-4.1 and NetMHCIIpan-4.0： improved predictions of MHC antigen presentation by concurrent motif deconvolution and integration of MS MHC eluted ligand data ［J］. Nucleic Acids Research， 2020， 48（W1）： W449-W454.
[32]	VITA R， MAHAJAN S， OVERTON J A， et al. The Immune Epitope Database （IEDB）： 2018 update ［J］. Nucleic Acids Research， 2019， 47（D1）： D339-D343.

序号	TGNovo网络层	输出维度
1	mass embedding	（400， 256）
2	intensity embedding	（400， 256）
3	Transformer Encoder×2	（400， 256）
4	Embedding	（Len， 256）
5	position embedding	（Len， 256）
6	Transformer Decoder×1	（Len， 256）
7	Conv1d×3	（Len， 400， 256）
8	Batch Norm1d×3	（Len， 400， 256）
9	GRU	（Len， 400， 256）
10	Max Pooling	（Len， 256）
11	Sum	（Len， 256）
12	Linear×3	（Len， 28）

序号	TGNovo网络层	输出维度
1	mass embedding	（400， 256）
2	intensity embedding	（400， 256）
3	Transformer Encoder×2	（400， 256）
4	Embedding	（Len， 256）
5	position embedding	（Len， 256）
6	Transformer Decoder×1	（Len， 256）
7	Conv1d×3	（Len， 400， 256）
8	Batch Norm1d×3	（Len， 400， 256）
9	GRU	（Len， 400， 256）
10	Max Pooling	（Len， 256）
11	Sum	（Len， 256）
12	Linear×3	（Len， 28）

参数	默认值
谱峰数	400
最大质荷比	6 000
最大肽段长度	100
束搜索最大值	10
Transformer编码器层数	2
Transformer解码器层数	1
Transformer注意力头数	4
隐藏层维度	256

参数	默认值
谱峰数	400
最大质荷比	6 000
最大肽段长度	100
束搜索最大值	10
Transformer编码器层数	2
Transformer解码器层数	1
Transformer注意力头数	4
隐藏层维度	256

数据集PID	物种	发表年份	谱图数	磷酸化谱图数	数据库
PXD008211	大肠杆菌	2018	8 722	1 471	大肠杆菌
PXD004452	人	2017	70 454	33 775	人
PXD019697	小鼠	2020	12 492	9 757	小鼠
PXD011284	拟南芥	2019	373 874	4 197	拟南芥
PXD023361	酵母菌	2021	20 044	146	酵母菌
PXD000138	合成	2013	46 885	46 885	合成肽
PXD007058	合成	2019	947	947	合成肽
PXD013210	合成	2019	12 118	12 118	合成肽
PXD004894	黑色素肿瘤	2017	4 101 232	未知	人

Transformer and gated recurrent unit-based de novo sequencing algorithm for phosphopeptides

基于Transformer和门控循环单元的磷酸化肽从头测序算法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 14

References 32

Related Articles 15

Recommended Articles

Metrics

参数	肽水平召回率/%	氨基酸水平召回率/%	单张谱图耗时/ms
默认值	83.2	94.3	133.3
谱峰数300	81.9	92.7	111.1
谱峰数500	83.4	94.7	142.8
编码器层数4	83.6	94.9	138.8
解码器层数2	84.1	95.3	142.8
注意力头数1	81.1	90.1	128.2
注意力头数8	82.9	92.4	133.3
束搜索宽度5	80.6	86.5	103.0
束搜索宽度20	83.1	94.2	166.6

实验设置	肽水平召回率	氨基酸水平召回率
TGNovo （完整模型）	83.2	94.3
移除GRU模块	66.1	86.4
移除Transformer模块	74.6	92.5
移除谱峰连接图	77.1	92.4
移除磷酸化修饰特征处理	72.5	91.2

[1]	Lifang WANG, Wenjing REN, Xiaodong GUO, Rongguo ZHANG, Lihua HU. Trident generative adversarial network for low-dose CT image denoising [J]. Journal of Computer Applications, 2026, 46(1): 270-279.
[2]	Junheng WU, Xiaodong WANG, Qixue HE. Time series prediction model based on statistical distribution sensing and frequency domain dual-channel fusion [J]. Journal of Computer Applications, 2026, 46(1): 113-123.
[3]	Yu SANG, Tong GONG, Chen ZHAO, Bowen YU, Siman LI. Domain-adaptive nighttime object detection method with photometric alignment [J]. Journal of Computer Applications, 2026, 46(1): 242-251.
[4]	Fang WANG, Jing HU, Rui ZHANG, Wenting FAN. Medical image segmentation network with content-guided multi-angle feature fusion [J]. Journal of Computer Applications, 2025, 45(9): 3017-3025.
[5]	Li LI, Han SONG, Peihe LIU, Hanlin CHEN. Named entity recognition for sensitive information based on data augmentation and residual networks [J]. Journal of Computer Applications, 2025, 45(9): 2790-2797.
[6]	Yiming LIANG, Jing FAN, Wenze CHAI. Multi-scale feature fusion sentiment classification based on bidirectional cross attention [J]. Journal of Computer Applications, 2025, 45(9): 2773-2782.
[7]	Jin LI, Liqun LIU. SAR and visible image fusion based on residual Swin Transformer [J]. Journal of Computer Applications, 2025, 45(9): 2949-2956.
[8]	Jinggang LYU, Shaorui PENG, Shuo GAO, Jin ZHOU. Speech enhancement network driven by complex frequency attention and multi-scale frequency enhancement [J]. Journal of Computer Applications, 2025, 45(9): 2957-2965.
[9]	Jing WANG, Jiaxing LIU, Wanying SONG, Jiaxing XUE, Wenxin DING. Few-shot skin image classification model based on spatial transformer network and feature distribution calibration [J]. Journal of Computer Applications, 2025, 45(8): 2720-2726.
[10]	Jin ZHOU, Yuzhi LI, Xu ZHANG, Shuo GAO, Li ZHANG, Jiachuan SHENG. Modulation recognition network for complex electromagnetic environments [J]. Journal of Computer Applications, 2025, 45(8): 2672-2682.
[11]	Yongpeng TAO, Shiqi BAI, Zhengwen ZHOU. Neural architecture search for multi-tissue segmentation using convolutional and transformer-based networks in glioma segmentation [J]. Journal of Computer Applications, 2025, 45(7): 2378-2386.
[12]	Haoyu LIU, Pengwei KONG, Yaoli WANG, Qing CHANG. Pedestrian detection algorithm based on multi-view information [J]. Journal of Computer Applications, 2025, 45(7): 2325-2332.
[13]	Dehui ZHOU, Jun ZHAO, Jinfeng CHENG. Tiny defect detection algorithm for bearing surface based on RT-DETR [J]. Journal of Computer Applications, 2025, 45(6): 1987-1997.
[14]	Sheping ZHAI, Yan HUANG, Qing YANG, Rui YANG. Multi-view entity alignment combining triples and text attributes [J]. Journal of Computer Applications, 2025, 45(6): 1793-1800.
[15]	Pengcheng XU, Lei HE, Chuan LI, Weiqi QIAN, Tun ZHAO. Deep symbolic regression method based on Transformer [J]. Journal of Computer Applications, 2025, 45(5): 1455-1463.