一种时序信号分类算法在纳米孔道离子电流信号识别中的应用★

doi:10.6023/A23040113

化学学报 ›› 2023, Vol. 81 ›› Issue (8): 912-919.DOI: 10.6023/A23040113 上一篇下一篇

所属专题：庆祝《化学学报》创刊90周年合辑

研究论文

一种时序信号分类算法在纳米孔道离子电流信号识别中的应用^★

倪雪^a, 辛凯莉^b, 胡正利^b^,^*(), 蒋翠玲^a^,^*(), 万永菁^a, 应佚伦^b, 龙亿涛^b

a 华东理工大学信息科学与工程学院上海 200237
b 南京大学化学化工学院分子传感与成像中心南京 210023

投稿日期:2023-04-03 发布日期:2023-09-14
作者简介:
^★庆祝《化学学报》创刊90周年.
† 共同第一作者
基金资助:
项目受科技部重点研发计划(2022YFA1304604); 国家自然科学基金(22106066); 国家自然科学基金(22027806)

A Time-Series Signal Classification Algorithm and Its Application to Nanopore Ionic Current Signal Identification^★

Xue Ni^a, Kaili Xin^b, Zhengli Hu^b(), Cuiling Jiang^a(), Yongjing Wan^a, Yi-Lun Ying^b, Yi-Tao Long^b

a School of Information Science and Engineering, East China University of Science and Technology, Shanghai 200237
b School of Chemistry and Chemical Engineering, Molecular Sensing and Imaging Center (MSIC), Nanjing University, Nanjing 210023

Received:2023-04-03 Published:2023-09-14
Contact: *E-mail: zhenglihu@nju.edu.cn; cuilingjiang@ecust.edu.cn
About author:
^★Dedicated to the 90th anniversary of Acta Chimica Sinica.
† These authors contributed equally to this work.
Supported by:
Ministry of Science and Technology Key R&D Program of China(2022YFA1304604); National Natural Science Foundation of China(22106066); National Natural Science Foundation of China(22027806)

文章分享

1. Supporting Information-A23040113.pdf(1166KB)

摘要/Abstract

纳米孔道单分子分析技术通常利用阻断电流的时域特征进行识别. 但对于结构、分子量等相似的物质, 由于其时域特性交叠, 采用传统的纳米孔道识别方法难以准确分辨. 为了充分挖掘具有差异性的深层特征, 提升纳米孔道离子流信号识别准确率, 提出了一种时序信号分类算法. 通过有重叠的滑动窗口对原始信号进行分帧, 并利用连续小波变换对逐帧信号进行处理, 可准确获取单分子事件的时频域浅层特征信息. 在此基础上, 利用多分支层间特征融合网络处理获取深层特征. 采用可信统计预测策略对子信号的分类概率统计, 该算法对单氨基酸差异多肽的纳米孔道离子电流信号的识别准确率高达99.00%, 可显著提高纳米孔道对分子量相似甚至相同的单分子的传感能力.

关键词: 纳米孔道分析, 时序信号, 深层特征提取, 层间特征融合, 可信统计预测

Nanopore-based single molecular analysis technique usually uses time-domain features such as time-current scatter plots of blocking currents for event recognition. However, as the time-domain features overlap with each other, the substances with extremely similar molecular structures are difficult to be accurately discriminated using traditional nanopore recognition methods. The differences in the deep feature representations need fully explored to obtain credible recognition results, thus improving the recognition accuracy of nanopore ionic current signals. Here, a time-series signal classification algorithm is proposed in this paper: firstly, the original signal is framed with overlapping sliding windows to generate sub-signals and extract their shallow feature information; then a time-series signal classification network based on Emphasized Channel Attention, Propagation and Aggregation in time delay neural network (ECAPA-TDNN) is proposed to develop a multi-branch inter-layer feature fusion model for deep feature extraction, where the multi-branch multi-level attention module of this model (RepVGG-SE-Res2Block, RSR-Block) obtains multi-scale features by constructing a feature pyramid structure within each residual block, reduces the inference speed based on structural reparameterization techniques while ensuring the model performance, and introduces Adaptively Spatial Feature Fusion (ASFF) to fuse the features of different layers in the network; finally, a credible statistical prediction strategy is used to obtain reliable classification results by counting the classification probabilities of sub-signals. The experimental results show that for the peptide sequences N'-DDFFIFFDD-C' (DF_I) and N'-DDFFLFFDD- C' (DF_L) containing only the different amino acids I (isoleucine) and L (leucine), which are isomers of each other, the algorithm achieves a recognition accuracy of 99.00%, obviously improving the sensing capability of nanopores for single molecules with similar or even identical molecular weights.

Key words: nanopore analysis, time-series signal, deep feature extraction, inter-layer feature fusion, credible statistical predictions

引用此文

倪雪, 辛凯莉, 胡正利, 蒋翠玲, 万永菁, 应佚伦, 龙亿涛. 一种时序信号分类算法在纳米孔道离子电流信号识别中的应用^★[J]. 化学学报, 2023, 81(8): 912-919.

Xue Ni, Kaili Xin, Zhengli Hu, Cuiling Jiang, Yongjing Wan, Yi-Lun Ying, Yi-Tao Long. A Time-Series Signal Classification Algorithm and Its Application to Nanopore Ionic Current Signal Identification^★[J]. Acta Chimica Sinica, 2023, 81(8): 912-919.

导出引用 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

分享此文

图/表 6

图1. (a)突变型Aerolysin纳米孔道检测多肽的原理示意图. 多肽DF_L序列为N'-DDFFLFFDD-C', 多肽DF_I序列为N'-DDFFIFFDD-C'. 多肽DF_L与DF_I的原始离子电流信号图(b)和时间-电流散点图(c). I/I0表示单个多肽分子进入纳米孔道产生的残余电流程度, 其中I0为单个Aerolysin孔道产生的开孔电流, I为分子进入孔道后的残余电流, τoff为单个多肽分子在孔道中的停留时间. (d)基于多分支层间特征融合网络的纳米孔道离子电流识别算法流程图

Figure 1. (a) Mutant aerolysin nanopore detection of single peptides. The sequences of DF_L and DF_I peptides are N'-DDFFLFFDD-C' and N'-DDFFIFFDD-C', respectively. (b) Ionic current signals, and (c) time-current scatter plot of different peptide substances. Herein, I/I0 indicates the degree of residual current generated by a single peptide molecule entering the nanopore channel, where I0 is the opening current generated by a single Aerolysin pore, I is the residual current after the molecule enters the pore, and τoff represents the residence time of a single peptide molecule in the pore. (d) Flowchart of a nanopore ionic current signal recognition algorithm based on multi-branch interlayer feature fusion networks

图2. 多分支层间特征融合网络结构图

Figure 2. Structural diagram of multi-branch interlayer feature fusion network

l	r	ACC/%
256	1/4	89.00
256	1/2	90.18
256	3/4	91.61
192	3/4	90.89
128	3/4	90.17

表1. 子信号分帧参数对实验结果的影响

Table 1. Effect of sub-signal segmentation parameters on experimental results

Models	Subsignal results				Statistical results
Models	Recall/%	Precision/%	F1/%	ACC/%	Recall/%	Precision/%	F1/%	ACC/%
Backbone	90.18	90.37	90.25	90.29	98.00	98.00	98.00	98.00
Backbone with RepVGG	91.09	91.27	91.16	91.19	98.00	98.00	98.00	98.00
Backbone with ASFF	91.39	91.36	91.37	91.39	99.00	99.02	98.99	99.00
This work	91.61	91.58	91.59	91.61	99.00	99.02	98.99	99.00

表2. 消融实验结果

Table 2. Results of ablation experiments

图3. 消融实验ROC曲线图

Figure 3. ROC curves of ablation experiment

图4. t-SNE特征分布可视化

Figure 4. Visualization of t-SNE feature distribution

参考文献 32

[1]	Afshar Bakshloo M.; Kasianowicz J. J.; Pastoriza-Gallego M.; Mathé J.; Daniel R.; Piguet F.; Oukhaled A. J. Am. Chem. Soc. 2022, 144, 2716. doi: 10.1021/jacs.1c11758
[2]	Derrington I. M.; Butler T. Z.; Collins M. D.; Manrao E.; Pavlenok M.; Niederweis M.; Gundlach J. H. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 16060. doi: 10.1073/pnas.1001831107
[3]	Hu Z.-L.; Huo M.-Z.; Ying Y.-L.; Long Y.-T. Angew. Chem., Int. Ed. 2021, 60, 14738. doi: 10.1002/anie.v60.27
[4]	Pratanwanich P. N.; Yao F.; Chen Y.; Koh C. W.; Wan Y. K.; Hendra C.; Poon P.; Goh Y. T.; Yap P. M.; Chooi J. Y. Nat. Biotechnol. 2021, 39, 1394. doi: 10.1038/s41587-021-00949-w
[5]	Goenka S. D.; Gorzynski J. E.; Shafin K.; Fisk D. G.; Pesout T.; Jensen T. D.; Monlong J.; Chang P. C.; Baid G.; Bernstein J. A. Nat. Biotechnol. 2022, 40, 1035. doi: 10.1038/s41587-022-01221-5
[6]	Sheng Y.; Zhou K.; Liu L.; Wu H.-C. Angew. Chem., Int. Ed. 2022, 61, e202200866. doi: 10.1002/anie.v61.20
[7]	Ju Y.; Pu M.; Sun K.; Song G.; Geng J. Chem. - Asian J. 2022, 17, e202200774. doi: 10.1002/asia.v17.22
[8]	Zhang Z.; Li T.; Sheng Y.; Liu L.; Wu H.-C. Small 2019, 15, 1804078. doi: 10.1002/smll.v15.2
[9]	Tsutsui M.; Yokota K.; Yoshida T.; Hotehama C.; Kowada H.; Esaki Y.; Taniguchi M.; Washio T.; Kawai T. ACS Sens. 2019, 4, 748. doi: 10.1021/acssensors.9b00113
[10]	Tanimoto I. M. F.; Cressiot B.; Greive S. J.; Le Pioufle B.; Bacri L.; Pelta J. Nano Res. 2022, 15, 9906. doi: 10.1007/s12274-022-4379-2
[11]	Wu X. Y.; Ying Y. L.; Long Y. T. Chem. J. Chin. Univ. 2019, 40, 1825. (in Chinese)
	( 武雪原, 应佚伦, 龙亿涛, 高等学校化学学报, 2019, 40, 1825.)
[12]	Wang L. Y.; Wang H. W.; Lu S. M.; Long Y. T. Chin. J. Anal. Chem. 2021, 49, 1166. (in Chinese)
	( 王凌远, 王浩炜, 芦思珉, 龙亿涛, 分析化学, 2021, 49, 1166.)
[13]	Lu Y.; Hu Z. L.; Ying Y. L.; Long Y. T. J. Instrum. Anal. 2019, 38, 129. (in Chinese)
	( 路瑶, 胡正利, 应佚伦, 龙亿涛, 分析测试学报, 2019, 38, 129.)
[14]	Chen Z.; Wang Z.; Xu Y.; Zhang X.; Tian B.; Bai J. Chem. Sci. 2021, 12, 15750. doi: 10.1039/D1SC04342K
[15]	Zernia S.; van der Heide N. J.; Galenkamp N. S.; Gouridis G.; Maglia G. ACS Nano 2020, 14, 2296. doi: 10.1021/acsnano.9b09434
[16]	Niu H. Y.; Hu Z. L.; Ying Y. L.; Long Y. T. Acta Chim. Sinica 2019, 77, 989. (in Chinese) doi: 10.6023/A19060230
	( 牛红艳, 胡正利, 应佚伦, 龙亿涛, 化学学报, 2019, 77, 989.)
[17]	Li M. Y.; Ying Y. L.; Long Y. T. Acta Chim. Sinica 2019, 77, 984. (in Chinese) doi: 10.6023/A19060202
	( 李孟寅, 应佚伦, 龙亿涛, 化学学报, 2019, 77, 984.)
[18]	Huo M.-Z.; Li M.-Y.; Ying Y.-L.; Long Y.-T. Anal. Chem. 2021, 93, 11364. doi: 10.1021/acs.analchem.1c00851
[19]	Ouldali H.; Sarthak K.; Ensslen T.; Piguet F.; Manivet P.; Pelta J.; Behrends J. C.; Aksimentiev A.; Oukhaled A. Nat. Biotechnol. 2020, 38, 176. doi: 10.1038/s41587-019-0345-2
[20]	Xin K.-L.; Hu Z.-L.; Liu S.-C.; Li X.; Li J.-G.; Niu H.; Ying Y.-L.; Long Y.-T. Angew. Chem., Int. Ed. 2022, 61, e202209970. doi: 10.1002/anie.v61.44
[21]	Hu R.; Rodrigues J. V.; Waduge P.; Yamazaki H.; Cressiot B.; Chishti Y.; Makowski L.; Yu D.; Shakhnovich E.; Zhao Q. ACS Nano 2018, 12, 4494. doi: 10.1021/acsnano.8b00734
[22]	Liu Y.; Zhang S.; Wang Y.; Wang L.; Cao Z.; Sun W.; Fan P.; Zhang P.; Chen H. Y.; Huang S. J. Am. Chem. Soc. 2022, 144, 13717. doi: 10.1021/jacs.2c04595
[23]	Li S. J.; Li X.; Wan Y.-J.; Ying Y.-L.; Yu R.-J.; Long Y.-T. Chem. - Asian J. 2023, 18, e202201144. doi: 10.1002/asia.v18.3
[24]	Liu S.-C.; Li M.-X.; Li M.-Y.; Wang Y.-Q.; Ying Y.-L.; Wan Y.-J.; Long Y.-T. Faraday Discuss. 2018, 210, 87. doi: 10.1039/C8FD00023A
[25]	Li X.; Ying Y.-L.; Fu X.-X.; Wan Y.-J.; Long Y.-T. Angew. Chem., Int. Ed. 2021, 60, 24582. doi: 10.1002/anie.v60.46
[26]	Desplanques B.; Thienpondt J.; Demuynck K.In 21st Annual Conference of the International Speech Communication Association (INTERSPEECH 2020), Shanghai, 2020, pp. 3830-3834.
[27]	Morlet J.; Arens G.; Fourgeau E.; Giard D. Geophysics 1982, 47, 222. doi: 10.1190/1.1441329
[28]	Niu H.; Li M.-Y.; Ying Y.-L.; Long Y.-T. Chem. Sci. 2022, 13, 2456. doi: 10.1039/D1SC06459B
[29]	Cao C.; Li M.-Y.; Cirauqui N.; Wang Y.-Q.; Dal Peraro M.; Tian H.; Long Y.-T. Nat. Commun. 2018, 9, 2823. doi: 10.1038/s41467-018-05108-5
[30]	Fawcett T. Pattern Recognit. Lett. 2006, 27, 861. doi: 10.1016/j.patrec.2005.10.010
[31]	Van der Maaten L.; Hinton G. J. Mach. Learn. Res. 2008, 9, 2579.
[32]	Yang C. Y.; Gu Z.; Hu Z. L.; Ying Y. L.; Long Y. T. J. Electrochem. 2019, 25, 312. (in Chinese)
	( 杨铖宇, 顾震, 胡正利, 应佚伦, 龙亿涛, 电化学, 2019, 25, 312.)

一种时序信号分类算法在纳米孔道离子电流信号识别中的应用^★

A Time-Series Signal Classification Algorithm and Its Application to Nanopore Ionic Current Signal Identification^★

RichHTML

PDF

补充材料

可视化

摘要/Abstract

引用此文

分享此文

图/表 6

参考文献 32

相关文章 0

编辑推荐

相关统计

本文评价

一种时序信号分类算法在纳米孔道离子电流信号识别中的应用★

A Time-Series Signal Classification Algorithm and Its Application to Nanopore Ionic Current Signal Identification★

RichHTML

PDF

补充材料

可视化

摘要/Abstract

引用此文

分享此文

图/表 6

参考文献 32

相关文章 0

编辑推荐

相关统计

本文评价

一种时序信号分类算法在纳米孔道离子电流信号识别中的应用^★

A Time-Series Signal Classification Algorithm and Its Application to Nanopore Ionic Current Signal Identification^★