Zhen-Lin Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhen-Lin Chen has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Hao Chi
Institute of Computing Technology, CAS
3 shared publications - Meng-Qiu Dong
National Institute of Biological Sciences, Beijing
3 shared publications - Si-Min He
Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS), Institute of Computing Technology, CAS, Beijing, China.
3 shared publications Peng-Zhi Mao
Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS), Institute of Computing Technology, CAS, Beijing, 100190, China
2 shared publications- Yong Cao
Shanghai Institute of Immunity and Infection, Chinese Academy of Sciences, Shanghai, China
2 shared publications - Adalet Memetimin
Beijing Normal University 19 Xinjiekouwai Avenue
1 shared publication
Latest Publications
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 4501-4506
MMRM: A Multiplex Multimodal Representation Model for Product Ranking in E-commerce Search
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 4501-4506
Open AccessMMRM: A Multiplex Multimodal Representation Model for Product Ranking in E-commerce Search
- Open Access
Improved identification of peptides, modification sites, and cross-link sites by Target-enhanced Accurate Inclusion Mass Screening (TAIMS)
Chemical cross-linking of proteins coupled with mass spectrometry provides structural insights by identifying cross-linked peptide pairs, abbreviated as cross-links. Presently, cross-link identification suffers from ambiguity and poor sensitivity, because they are typically of lower abundance and consequently of lower MS2 quality than linear peptides present in the same sample. Here, we present Target-enhanced Accurate Inclusion Mass Screening (TAIMS), a meticulously optimized targeted mass spectrometry method. TAIMS significantly improved the quality of MS2 as indicated by fragment ion coverage (FIC) and other metrics. From data-dependent acquisition (DDA) to TAIMS, high-FIC cross-links increased by 401 or 678 on yeast ribosome or E. coli lysate, respectively, or from about 40% to around 90%. As a result, TAIMS substantially enhanced the accuracy of cross-link site localization and mitigated sensitivity loss in cross-link identification from large database searches. Enhanced identification sensitivity of TAIMS is further evidenced by its capacity to recover genuine cross-link identifications from data that would typically be discarded. On cross-linked E. coli lysate, 10.5% (284/2711) of the inclusion-list entries generated from unidentified cross-link-spectrum matches gained identity through TAIMS. Of these, 230 were linear peptides and 54 were cross-links, including 10 inter-molecular cross-links missed entirely by DDA. Additionally, we demonstrate that TAIMS is a general method for identification of low-abundance, post-translationally modified peptides. On a mouse brain sample, TAIMS increased the number of phosphopeptides identified with accurate phosphosite assignment by 67%. These findings indicate that TAIMS holds broad applicability in proteomics.
Open AccessImproved identification of peptides, modification sites, and cross-link sites by Target-enhanced Accurate Inclusion Mass Screening (TAIMS)
Chemical cross-linking of proteins coupled with mass spectrometry provides structural insights by identifying cross-linked peptide pairs, abbreviated as cross-links. Presently, cross-link identification suffers from ambiguity and poor sensitivity, because they are typically of lower abundance and consequently of lower MS2 quality than linear peptides present in the same sample. Here, we present Target-enhanced Accurate Inclusion Mass Screening (TAIMS), a meticulously optimized targeted mass spectrometry method. TAIMS significantly improved the quality of MS2 as indicated by fragment ion coverage (FIC) and other metrics. From data-dependent acquisition (DDA) to TAIMS, high-FIC cross-links increased by 401 or 678 on yeast ribosome or E. coli lysate, respectively, or from about 40% to around 90%. As a result, TAIMS substantially enhanced the accuracy of cross-link site localization and mitigated sensitivity loss in cross-link identification from large database searches. Enhanced identification sensitivity of TAIMS is further evidenced by its capacity to recover genuine cross-link identifications from data that would typically be discarded. On cross-linked E. coli lysate, 10.5% (284/2711) of the inclusion-list entries generated from unidentified cross-link-spectrum matches gained identity through TAIMS. Of these, 230 were linear peptides and 54 were cross-links, including 10 inter-molecular cross-links missed entirely by DDA. Additionally, we demonstrate that TAIMS is a general method for identification of low-abundance, post-translationally modified peptides. On a mouse brain sample, TAIMS increased the number of phosphopeptides identified with accurate phosphosite assignment by 67%. These findings indicate that TAIMS holds broad applicability in proteomics.
- Open Access
- 9 May 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 43 (4) , 1-25
Enhancing Sequential Personalized Product Search with External Out-of-sequence Knowledge
A key challenge in personalized product search is to capture user’s preferences. Recent work attempted to model sequences of user historical behaviors, i.e., product purchase histories, to build user profiles and to personalize results accordingly. Although these approaches have demonstrated promising retrieval performances, we notice that most of them focus solely on the intra-sequence interactions between items. However, as there is usually a small amount of historical behavior data, the user profiles learned by these approaches could be very sensitive to the noise included in it. To tackle this problem, we propose incorporating out-of-sequence external information to enhance user modeling. More specifically, we inject the external item–item relations (e.g., belonging to the same brand), and query–query relations (e.g., the semantic similarities between them), into the intra-sequence interaction to learn better user profiles. In addition, we devise two auxiliary decoders, with the historical item sequence reconstruction task and the global item similarity prediction task, to further improve the reliability of user modeling. Experimental results on two datasets from simulated and real user search logs respectively show that the proposed personalized product search method outperforms existing approaches.
- 9 May 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 43 (4) , 1-25
Open AccessEnhancing Sequential Personalized Product Search with External Out-of-sequence Knowledge
A key challenge in personalized product search is to capture user’s preferences. Recent work attempted to model sequences of user historical behaviors, i.e., product purchase histories, to build user profiles and to personalize results accordingly. Although these approaches have demonstrated promising retrieval performances, we notice that most of them focus solely on the intra-sequence interactions between items. However, as there is usually a small amount of historical behavior data, the user profiles learned by these approaches could be very sensitive to the noise included in it. To tackle this problem, we propose incorporating out-of-sequence external information to enhance user modeling. More specifically, we inject the external item–item relations (e.g., belonging to the same brand), and query–query relations (e.g., the semantic similarities between them), into the intra-sequence interaction to learn better user profiles. In addition, we devise two auxiliary decoders, with the historical item sequence reconstruction task and the global item similarity prediction task, to further improve the reliability of user modeling. Experimental results on two datasets from simulated and real user search logs respectively show that the proposed personalized product search method outperforms existing approaches.
- Open Access
- 18 April 2025
- Published by Elsevier in Molecular & Cellular Proteomics
- Vol. 24 (7) , 100975
Global Analysis of Protein and Small-Molecule Substrates of Ubiquitin-Like Proteins
- 18 April 2025
- Published by Elsevier in Molecular & Cellular Proteomics
- Vol. 24 (7) , 100975
Open AccessGlobal Analysis of Protein and Small-Molecule Substrates of Ubiquitin-Like Proteins
- Top Cited PapersOpen Access
- 26 October 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-19
Predicting transcriptional responses to novel chemical perturbations using deep generative model for drug discovery
Understanding transcriptional responses to chemical perturbations is central to drug discovery, but exhaustive experimental screening of disease-compound combinations is unfeasible. To overcome this limitation, here we introduce PRnet, a perturbation-conditioned deep generative model that predicts transcriptional responses to novel chemical perturbations that have never experimentally perturbed at bulk and single-cell levels. Evaluations indicate that PRnet outperforms alternative methods in predicting responses across novel compounds, pathways, and cell lines. PRnet enables gene-level response interpretation and in-silico drug screening for diseases based on gene signatures. PRnet further identifies and experimentally validates novel compound candidates against small cell lung cancer and colorectal cancer. Lastly, PRnet generates a large-scale integration atlas of perturbation profiles, covering 88 cell lines, 52 tissues, and various compound libraries. PRnet provides a robust and scalable candidate recommendation workflow and successfully recommends drug candidates for 233 diseases. Overall, PRnet is an effective and valuable tool for gene-based therapeutics screening.
- 26 October 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-19
Top Cited PapersOpen AccessPredicting transcriptional responses to novel chemical perturbations using deep generative model for drug discovery
Understanding transcriptional responses to chemical perturbations is central to drug discovery, but exhaustive experimental screening of disease-compound combinations is unfeasible. To overcome this limitation, here we introduce PRnet, a perturbation-conditioned deep generative model that predicts transcriptional responses to novel chemical perturbations that have never experimentally perturbed at bulk and single-cell levels. Evaluations indicate that PRnet outperforms alternative methods in predicting responses across novel compounds, pathways, and cell lines. PRnet enables gene-level response interpretation and in-silico drug screening for diseases based on gene signatures. PRnet further identifies and experimentally validates novel compound candidates against small cell lung cancer and colorectal cancer. Lastly, PRnet generates a large-scale integration atlas of perturbation profiles, covering 88 cell lines, 52 tissues, and various compound libraries. PRnet provides a robust and scalable candidate recommendation workflow and successfully recommends drug candidates for 233 diseases. Overall, PRnet is an effective and valuable tool for gene-based therapeutics screening.
- 6 April 2021
- Published by American Chemical Society (ACS) in Journal of Proteome Research
- Vol. 20 (5) , 2570-2582
pDeepXL: MS/MS Spectrum Prediction for Cross-Linked Peptide Pairs by Deep Learning
In cross-linking mass spectrometry, the identification of cross-linked peptide pairs heavily relies on the ability of a database search engine to measure the similarities between experimental and theoretical MS/MS spectra. However, the lack of accurate ion intensities in theoretical spectra impairs the performance of search engines, in particular, on proteome scales. Here we introduce pDeepXL, a deep neural network to predict MS/MS spectra of cross-linked peptide pairs. To train pDeepXL, we used the transfer-learning technique because it facilitated the training with limited benchmark data of cross-linked peptide pairs. Test results on more than ten data sets showed that pDeepXL accurately predicted the spectra of both noncleavable DSS/BS3/Leiker cross-linked peptide pairs (>80% of predicted spectra have Pearson’s r values higher than 0.9) and cleavable DSSO/DSBU cross-linked peptide pairs (>75% of predicted spectra have Pearson’s r values higher than 0.9). pDeepXL also achieved the accurate prediction on unseen data sets using an online fine-tuning technique. Lastly, integrating pDeepXL into a database search engine increased the number of identified cross-link spectra by 18% on average.
- 6 April 2021
- Published by American Chemical Society (ACS) in Journal of Proteome Research
- Vol. 20 (5) , 2570-2582
pDeepXL: MS/MS Spectrum Prediction for Cross-Linked Peptide Pairs by Deep Learning
In cross-linking mass spectrometry, the identification of cross-linked peptide pairs heavily relies on the ability of a database search engine to measure the similarities between experimental and theoretical MS/MS spectra. However, the lack of accurate ion intensities in theoretical spectra impairs the performance of search engines, in particular, on proteome scales. Here we introduce pDeepXL, a deep neural network to predict MS/MS spectra of cross-linked peptide pairs. To train pDeepXL, we used the transfer-learning technique because it facilitated the training with limited benchmark data of cross-linked peptide pairs. Test results on more than ten data sets showed that pDeepXL accurately predicted the spectra of both noncleavable DSS/BS3/Leiker cross-linked peptide pairs (>80% of predicted spectra have Pearson’s r values higher than 0.9) and cleavable DSSO/DSBU cross-linked peptide pairs (>75% of predicted spectra have Pearson’s r values higher than 0.9). pDeepXL also achieved the accurate prediction on unseen data sets using an online fine-tuning technique. Lastly, integrating pDeepXL into a database search engine increased the number of identified cross-link spectra by 18% on average.
- Top Cited PapersOpen Access
- 30 July 2019
- Published by Springer Nature in Nature Communications
- Vol. 10 (1) , 1-12
A high-speed search engine pLink 2 with systematic evaluation for proteome-scale identification of cross-linked peptides
We describe pLink 2, a search engine with higher speed and reliability for proteome-scale identification of cross-linked peptides. With a two-stage open search strategy facilitated by fragment indexing, pLink 2 is ~40 times faster than pLink 1 and 3~10 times faster than Kojak. Furthermore, using simulated datasets, synthetic datasets, 15N metabolically labeled datasets, and entrapment databases, four analysis methods were designed to evaluate the credibility of ten state-of-the-art search engines. This systematic evaluation shows that pLink 2 outperforms these methods in precision and sensitivity, especially at proteome scales. Lastly, re-analysis of four published proteome-scale cross-linking datasets with pLink 2 required only a fraction of the time used by pLink 1, with up to 27% more cross-linked residue pairs identified. pLink 2 is therefore an efficient and reliable tool for cross-linking mass spectrometry analysis, and the systematic evaluation methods described here will be useful for future software development.
- 30 July 2019
- Published by Springer Nature in Nature Communications
- Vol. 10 (1) , 1-12
Top Cited PapersOpen AccessA high-speed search engine pLink 2 with systematic evaluation for proteome-scale identification of cross-linked peptides
We describe pLink 2, a search engine with higher speed and reliability for proteome-scale identification of cross-linked peptides. With a two-stage open search strategy facilitated by fragment indexing, pLink 2 is ~40 times faster than pLink 1 and 3~10 times faster than Kojak. Furthermore, using simulated datasets, synthetic datasets, 15N metabolically labeled datasets, and entrapment databases, four analysis methods were designed to evaluate the credibility of ten state-of-the-art search engines. This systematic evaluation shows that pLink 2 outperforms these methods in precision and sensitivity, especially at proteome scales. Lastly, re-analysis of four published proteome-scale cross-linking datasets with pLink 2 required only a fraction of the time used by pLink 1, with up to 27% more cross-linked residue pairs identified. pLink 2 is therefore an efficient and reliable tool for cross-linking mass spectrometry analysis, and the systematic evaluation methods described here will be useful for future software development.