Highly accurate and large-scale collision cross sections prediction with graph neural networks
Abstract The collision cross section (CCS) values derived from ion mobility spectrometry can be used to improve the accuracy of compound identification. Here, we have developed the Structure included graph merging with adduct method for CCS prediction (SigmaCCS) based on graph neural networks using...
Main Authors: | , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2023-07-01
|
Series: | Communications Chemistry |
Online Access: | https://doi.org/10.1038/s42004-023-00939-w |
_version_ | 1797784797339189248 |
---|---|
author | Renfeng Guo Youjia Zhang Yuxuan Liao Qiong Yang Ting Xie Xiaqiong Fan Zhonglong Lin Yi Chen Hongmei Lu Zhimin Zhang |
author_facet | Renfeng Guo Youjia Zhang Yuxuan Liao Qiong Yang Ting Xie Xiaqiong Fan Zhonglong Lin Yi Chen Hongmei Lu Zhimin Zhang |
author_sort | Renfeng Guo |
collection | DOAJ |
description | Abstract The collision cross section (CCS) values derived from ion mobility spectrometry can be used to improve the accuracy of compound identification. Here, we have developed the Structure included graph merging with adduct method for CCS prediction (SigmaCCS) based on graph neural networks using 3D conformers as inputs. A model was trained, evaluated, and tested with >5,000 experimental CCS values. It achieved a coefficient of determination of 0.9945 and a median relative error of 1.1751% on the test set. The model-agnostic interpretation method and the visualization of the learned representations were used to investigate the chemical rationality of SigmaCCS. An in-silico database with 282 million CCS values was generated for three different adduct types of 94 million compounds. Its source code is publicly available at https://github.com/zmzhang/SigmaCCS . Altogether, SigmaCCS is an accurate, rational, and off-the-shelf method to directly predict CCS values from molecular structures. |
first_indexed | 2024-03-13T00:45:00Z |
format | Article |
id | doaj.art-e344046a1a1a4948b0511cf9ad01be85 |
institution | Directory Open Access Journal |
issn | 2399-3669 |
language | English |
last_indexed | 2024-03-13T00:45:00Z |
publishDate | 2023-07-01 |
publisher | Nature Portfolio |
record_format | Article |
series | Communications Chemistry |
spelling | doaj.art-e344046a1a1a4948b0511cf9ad01be852023-07-09T11:08:27ZengNature PortfolioCommunications Chemistry2399-36692023-07-016111010.1038/s42004-023-00939-wHighly accurate and large-scale collision cross sections prediction with graph neural networksRenfeng Guo0Youjia Zhang1Yuxuan Liao2Qiong Yang3Ting Xie4Xiaqiong Fan5Zhonglong Lin6Yi Chen7Hongmei Lu8Zhimin Zhang9College of Chemistry and Chemical Engineering, Central South UniversitySchool of Computer Science and Technology, Huazhong University of Science and TechnologyCollege of Chemistry and Chemical Engineering, Central South UniversityCollege of Chemistry and Chemical Engineering, Central South UniversityCollege of Chemistry and Chemical Engineering, Central South UniversityCollege of Chemistry and Chemical Engineering, Central South UniversityYunnan Academy of Tobacco Agricultural SciencesYunnan Academy of Tobacco Agricultural SciencesCollege of Chemistry and Chemical Engineering, Central South UniversityCollege of Chemistry and Chemical Engineering, Central South UniversityAbstract The collision cross section (CCS) values derived from ion mobility spectrometry can be used to improve the accuracy of compound identification. Here, we have developed the Structure included graph merging with adduct method for CCS prediction (SigmaCCS) based on graph neural networks using 3D conformers as inputs. A model was trained, evaluated, and tested with >5,000 experimental CCS values. It achieved a coefficient of determination of 0.9945 and a median relative error of 1.1751% on the test set. The model-agnostic interpretation method and the visualization of the learned representations were used to investigate the chemical rationality of SigmaCCS. An in-silico database with 282 million CCS values was generated for three different adduct types of 94 million compounds. Its source code is publicly available at https://github.com/zmzhang/SigmaCCS . Altogether, SigmaCCS is an accurate, rational, and off-the-shelf method to directly predict CCS values from molecular structures.https://doi.org/10.1038/s42004-023-00939-w |
spellingShingle | Renfeng Guo Youjia Zhang Yuxuan Liao Qiong Yang Ting Xie Xiaqiong Fan Zhonglong Lin Yi Chen Hongmei Lu Zhimin Zhang Highly accurate and large-scale collision cross sections prediction with graph neural networks Communications Chemistry |
title | Highly accurate and large-scale collision cross sections prediction with graph neural networks |
title_full | Highly accurate and large-scale collision cross sections prediction with graph neural networks |
title_fullStr | Highly accurate and large-scale collision cross sections prediction with graph neural networks |
title_full_unstemmed | Highly accurate and large-scale collision cross sections prediction with graph neural networks |
title_short | Highly accurate and large-scale collision cross sections prediction with graph neural networks |
title_sort | highly accurate and large scale collision cross sections prediction with graph neural networks |
url | https://doi.org/10.1038/s42004-023-00939-w |
work_keys_str_mv | AT renfengguo highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT youjiazhang highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT yuxuanliao highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT qiongyang highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT tingxie highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT xiaqiongfan highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT zhonglonglin highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT yichen highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT hongmeilu highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks AT zhiminzhang highlyaccurateandlargescalecollisioncrosssectionspredictionwithgraphneuralnetworks |