Fei Teng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Fei Teng has a total of 166 co-authors. Here are the top co-authors by number of shared publications:
Ao Chen
BGI Research
3 shared publicationsSha Liao
BGI Research
3 shared publicationsAlicia Alonso
Department of Physiology and Biophysics, Weill Cornell Medicine, New York, USA
2 shared publicationsCharles M. Nicolet
Keck School of Medicine, University of Southern California, Los Angeles, USA
2 shared publications- Christopher E. Mason
Department of Physiology and Biophysics, Weill Cornell Medicine, New York, USA
2 shared publications Derek Warner
DNA Sequencing Core, University of Utah, Salt Lake City, USA
2 shared publications
Latest Publications
- Open Access
StereoMM: A Graph Fusion Model for Integrating Spatial Transcriptomic Data and Pathological Images
Spatially resolved omics technologies generating multimodal and high-throughput data lead to the urgent need for advanced analysis to allow the biological discoveries by comprehensively utilizing information from multi-omics data. The H&E image and spatial transcriptomic data indicate abundant features which are different and complementary to each other. AI algorithms can perform nonlinear analysis on these aligned or unaligned complex datasets to decode tumoral heterogeneity for detecting functional domain. However,the interpretability of AI-generated outcomes for human experts is a problem hindering application of multi-modal analysis in clinic. We presented a machine learning based toolchain called StereoMM, which is a graph fusion model that can integrate gene expression, histological images, and spatial location. StereoMM firstly performs information interaction on transcriptomic and imaging features through the attention module, guaranteeing explanations for its decision-making processes. The interactive features are input into the graph autoencoder together with the graph of spatial position, so that multimodal features are fused in a self-supervised manner. Here, StereoMM was subjected to mouse brain tissue, demonstrating its capability to discern fine tissue architecture, while highlighting its advantage in computational speed. Utilizing data from Stereo-seq of human lung adenosquamous carcinoma and 10X Visium of human breast cancer, we showed its superior performance in spatial domain recognition over competing software and its ability to reveal tumor heterogeneity. The fusion approach for imaging and gene expression data within StereoMM aids in the more accurate identification of domains, unveils critical molecular features, and elucidates the connections between different domains, thereby laying the groundwork for downstream analysis.
Open AccessStereoMM: A Graph Fusion Model for Integrating Spatial Transcriptomic Data and Pathological Images
Spatially resolved omics technologies generating multimodal and high-throughput data lead to the urgent need for advanced analysis to allow the biological discoveries by comprehensively utilizing information from multi-omics data. The H&E image and spatial transcriptomic data indicate abundant features which are different and complementary to each other. AI algorithms can perform nonlinear analysis on these aligned or unaligned complex datasets to decode tumoral heterogeneity for detecting functional domain. However,the interpretability of AI-generated outcomes for human experts is a problem hindering application of multi-modal analysis in clinic. We presented a machine learning based toolchain called StereoMM, which is a graph fusion model that can integrate gene expression, histological images, and spatial location. StereoMM firstly performs information interaction on transcriptomic and imaging features through the attention module, guaranteeing explanations for its decision-making processes. The interactive features are input into the graph autoencoder together with the graph of spatial position, so that multimodal features are fused in a self-supervised manner. Here, StereoMM was subjected to mouse brain tissue, demonstrating its capability to discern fine tissue architecture, while highlighting its advantage in computational speed. Utilizing data from Stereo-seq of human lung adenosquamous carcinoma and 10X Visium of human breast cancer, we showed its superior performance in spatial domain recognition over competing software and its ability to reveal tumor heterogeneity. The fusion approach for imaging and gene expression data within StereoMM aids in the more accurate identification of domains, unveils critical molecular features, and elucidates the connections between different domains, thereby laying the groundwork for downstream analysis.
- Open Access
- 2 January 2024
- Published by Oxford University Press (OUP) in GigaScience
- Vol. 13
StereoSiTE: a framework to spatially and quantitatively profile the cellular neighborhood organized iTME
Background: Spatial transcriptome (ST) technologies are emerging as powerful tools for studying tumor biology. However, existing tools for analyzing ST data are limited, as they mainly rely on algorithms developed for single-cell RNA sequencing data and do not fully utilize the spatial information. While some algorithms have been developed for ST data, they are often designed for specific tasks, lacking a comprehensive analytical framework for leveraging spatial information.Results: In this study, we present StereoSiTE, an analytical framework that combines open-source bioinformatics tools with custom algorithms to accurately infer the functional spatial cell interaction intensity (SCII) within the cellular neighborhood (CN) of interest. We applied StereoSiTE to decode ST datasets from xenograft models and found that the CN efficiently distinguished different cellular contexts, while the SCII analysis provided more precise insights into intercellular interactions by incorporating spatial information. By applying StereoSiTE to multiple samples, we successfully identified a CN region dominated by neutrophils, suggesting their potential role in remodeling the immune tumor microenvironment (iTME) after treatment. Moreover, the SCII analysis within the CN region revealed neutrophil-mediated communication, supported by pathway enrichment, transcription factor regulon activities, and protein–protein interactions.Conclusions: StereoSiTE represents a promising framework for unraveling the mechanisms underlying treatment response within the iTME by leveraging CN-based tissue domain identification and SCII-inferred spatial intercellular interactions. The software is designed to be scalable, modular, and user-friendly, making it accessible to a wide range of researchers.
- 2 January 2024
- Published by Oxford University Press (OUP) in GigaScience
- Vol. 13
Open AccessStereoSiTE: a framework to spatially and quantitatively profile the cellular neighborhood organized iTME
Background: Spatial transcriptome (ST) technologies are emerging as powerful tools for studying tumor biology. However, existing tools for analyzing ST data are limited, as they mainly rely on algorithms developed for single-cell RNA sequencing data and do not fully utilize the spatial information. While some algorithms have been developed for ST data, they are often designed for specific tasks, lacking a comprehensive analytical framework for leveraging spatial information.Results: In this study, we present StereoSiTE, an analytical framework that combines open-source bioinformatics tools with custom algorithms to accurately infer the functional spatial cell interaction intensity (SCII) within the cellular neighborhood (CN) of interest. We applied StereoSiTE to decode ST datasets from xenograft models and found that the CN efficiently distinguished different cellular contexts, while the SCII analysis provided more precise insights into intercellular interactions by incorporating spatial information. By applying StereoSiTE to multiple samples, we successfully identified a CN region dominated by neutrophils, suggesting their potential role in remodeling the immune tumor microenvironment (iTME) after treatment. Moreover, the SCII analysis within the CN region revealed neutrophil-mediated communication, supported by pathway enrichment, transcription factor regulon activities, and protein–protein interactions.Conclusions: StereoSiTE represents a promising framework for unraveling the mechanisms underlying treatment response within the iTME by leveraging CN-based tissue domain identification and SCII-inferred spatial intercellular interactions. The software is designed to be scalable, modular, and user-friendly, making it accessible to a wide range of researchers.
- Open Access
- 26 August 2023
- Published by Springer Nature in npj Precision Oncology
- Vol. 7 (1) , 1-14
Clonal dynamics and Stereo-seq resolve origin and phenotypic plasticity of adenosquamous carcinoma
The genomic origin and development of the biphasic lung adenosquamous carcinoma (ASC) remain inconclusive. Here, we derived potential evolutionary trajectory of ASC through whole-exome sequencing, Stereo-seq, and patient-derived xenografts. We showed that EGFR and MET activating mutations were the main drivers in ASCs. Phylogenetically, these drivers and passenger mutations found in both components were trunk clonal events, confirming monoclonal origination. Comparison of multiple lesions also revealed closer genomic distance between lymph node metastases and the ASC component with the same phenotype. However, as mutational signatures of EGFR-positive lung squamous carcinomas (LUSCs) were more comparable to EGFR-positive ASCs than to wild-type LUSCs, we postulated different origination of these LUSCs, with ASC being the potential intermediate state of driver-positive LUSCs. Spatial transcriptomic profiling inferred transformation from adenocarcinoma to squamous cell carcinoma, which was then histologically captured in vivo. Together, our results explained the development of ASC and provided insights into future clinical decisions.
- 26 August 2023
- Published by Springer Nature in npj Precision Oncology
- Vol. 7 (1) , 1-14
Open AccessClonal dynamics and Stereo-seq resolve origin and phenotypic plasticity of adenosquamous carcinoma
The genomic origin and development of the biphasic lung adenosquamous carcinoma (ASC) remain inconclusive. Here, we derived potential evolutionary trajectory of ASC through whole-exome sequencing, Stereo-seq, and patient-derived xenografts. We showed that EGFR and MET activating mutations were the main drivers in ASCs. Phylogenetically, these drivers and passenger mutations found in both components were trunk clonal events, confirming monoclonal origination. Comparison of multiple lesions also revealed closer genomic distance between lymph node metastases and the ASC component with the same phenotype. However, as mutational signatures of EGFR-positive lung squamous carcinomas (LUSCs) were more comparable to EGFR-positive ASCs than to wild-type LUSCs, we postulated different origination of these LUSCs, with ASC being the potential intermediate state of driver-positive LUSCs. Spatial transcriptomic profiling inferred transformation from adenocarcinoma to squamous cell carcinoma, which was then histologically captured in vivo. Together, our results explained the development of ASC and provided insights into future clinical decisions.
- Open Access
CellBin: a highly accurate single-cell gene expression processing pipeline for high-resolution spatial transcriptomics
Background: Owing to recent advances in resolution and field-of-view, spatially resolved transcriptomics sequencing, such as Stereo-seq, has emerged as a cutting-edge technology for the interpretation of large tissues at the single-cell level. To generate accurate single-cell spatial gene expression profiles from high-resolution spatial omics data, a powerful computational tool is required. Findings: We present CellBin, an image-facilitated one-stop pipeline for high-resolution and large field-of-view spatial transcriptomic data of Stereo-seq. CellBin provides a comprehensive and systematic platform for generating high-confidence single-cell spatial gene expression profiles, which specifically includes image stitching, image registration, tissue segmentation, nuclei segmentation and molecule labeling. CellBin is user-friendly and does not require a specific level of omics and image analysis expertise. Conclusions: During image stitching and molecule labeling, CellBin delivers better-performing algorithms to reduce stitching error and time, in addition to improving the signal-to-noise ratio of single-cell gene expression data, in comparison with existing methods. Additionally, CellBin has been shown to obtain highly accurate single-cell spatial data using mouse brain tissue, which facilitated clustering and annotation.
Open AccessCellBin: a highly accurate single-cell gene expression processing pipeline for high-resolution spatial transcriptomics
Background: Owing to recent advances in resolution and field-of-view, spatially resolved transcriptomics sequencing, such as Stereo-seq, has emerged as a cutting-edge technology for the interpretation of large tissues at the single-cell level. To generate accurate single-cell spatial gene expression profiles from high-resolution spatial omics data, a powerful computational tool is required. Findings: We present CellBin, an image-facilitated one-stop pipeline for high-resolution and large field-of-view spatial transcriptomic data of Stereo-seq. CellBin provides a comprehensive and systematic platform for generating high-confidence single-cell spatial gene expression profiles, which specifically includes image stitching, image registration, tissue segmentation, nuclei segmentation and molecule labeling. CellBin is user-friendly and does not require a specific level of omics and image analysis expertise. Conclusions: During image stitching and molecule labeling, CellBin delivers better-performing algorithms to reduce stitching error and time, in addition to improving the signal-to-noise ratio of single-cell gene expression data, in comparison with existing methods. Additionally, CellBin has been shown to obtain highly accurate single-cell spatial data using mouse brain tissue, which facilitated clustering and annotation.
- 11 October 2021
- Published by Springer Nature in Nature Biotechnology
- Vol. 39 (11) , 1466
Author Correction: Performance assessment of DNA sequencing platforms in the ABRF Next-Generation Sequencing Study
- 11 October 2021
- Published by Springer Nature in Nature Biotechnology
- Vol. 39 (11) , 1466
Author Correction: Performance assessment of DNA sequencing platforms in the ABRF Next-Generation Sequencing Study
- 9 September 2021
- Published by Springer Nature in Nature Biotechnology
- Vol. 39 (9) , 1129-1140
Performance assessment of DNA sequencing platforms in the ABRF Next-Generation Sequencing Study
Assessing the reproducibility, accuracy and utility of massively parallel DNA sequencing platforms remains an ongoing challenge. Here the Association of Biomolecular Resource Facilities (ABRF) Next-Generation Sequencing Study benchmarks the performance of a set of sequencing instruments (HiSeq/NovaSeq/paired-end 2 × 250-bp chemistry, Ion S5/Proton, PacBio circular consensus sequencing (CCS), Oxford Nanopore Technologies PromethION/MinION, BGISEQ-500/MGISEQ-2000 and GS111) on human and bacterial reference DNA samples. Among short-read instruments, HiSeq 4000 and X10 provided the most consistent, highest genome coverage, while BGI/MGISEQ provided the lowest sequencing error rates. The long-read instrument PacBio CCS had the highest reference-based mapping rate and lowest non-mapping rate. The two long-read platforms PacBio CCS and PromethION/MinION showed the best sequence mapping in repeat-rich areas and across homopolymers. NovaSeq 6000 using 2 × 250-bp read chemistry was the most robust instrument for capturing known insertion/deletion events. This study serves as a benchmark for current genomics technologies, as well as a resource to inform experimental design and next-generation sequencing variant calling.
- 9 September 2021
- Published by Springer Nature in Nature Biotechnology
- Vol. 39 (9) , 1129-1140
Performance assessment of DNA sequencing platforms in the ABRF Next-Generation Sequencing Study
Assessing the reproducibility, accuracy and utility of massively parallel DNA sequencing platforms remains an ongoing challenge. Here the Association of Biomolecular Resource Facilities (ABRF) Next-Generation Sequencing Study benchmarks the performance of a set of sequencing instruments (HiSeq/NovaSeq/paired-end 2 × 250-bp chemistry, Ion S5/Proton, PacBio circular consensus sequencing (CCS), Oxford Nanopore Technologies PromethION/MinION, BGISEQ-500/MGISEQ-2000 and GS111) on human and bacterial reference DNA samples. Among short-read instruments, HiSeq 4000 and X10 provided the most consistent, highest genome coverage, while BGI/MGISEQ provided the lowest sequencing error rates. The long-read instrument PacBio CCS had the highest reference-based mapping rate and lowest non-mapping rate. The two long-read platforms PacBio CCS and PromethION/MinION showed the best sequence mapping in repeat-rich areas and across homopolymers. NovaSeq 6000 using 2 × 250-bp read chemistry was the most robust instrument for capturing known insertion/deletion events. This study serves as a benchmark for current genomics technologies, as well as a resource to inform experimental design and next-generation sequencing variant calling.
- Open Access
- 1 September 2021
- Published by Springer Nature in Signal Transduction and Targeted Therapy
- Vol. 6 (1) , 1-10
Detection of an anti-angina therapeutic module in the effective population treated by a multi-target drug Danhong injection: a randomized trial
It’s a challenge for detecting the therapeutic targets of a polypharmacological drug from variations in the responsed networks in the differentiated populations with complex diseases, as stable coronary heart disease. Here, in an adaptive, 31-center, randomized, double-blind trial involving 920 patients with moderate symptomatic stable angina treated by 14-day Danhong injection(DHI), a kind of polypharmacological drug with high quality control, or placebo (0.9% saline), with 76-day following-up, we firstly confirmed that DHI could increase the proportion of patients with clinically significant changes on angina-frequency assessed by Seattle Angina Questionnaire (ΔSAQ-AF ≥ 20) (12.78% at Day 30, 95% confidence interval [CI] 5.86–19.71%, P = 0.0003, 13.82% at Day 60, 95% CI 6.82–20.82%, P = 0.0001 and 8.95% at Day 90, 95% CI 2.06–15.85%, P = 0.01). We also found that there were no significant differences in new-onset major vascular events (P = 0.8502) and serious adverse events (P = 0.9105) between DHI and placebo. After performing the RNA sequencing in 62 selected patients, we developed a systemic modular approach to identify differentially expressed modules (DEMs) of DHI with the Zsummary value less than 0 compared with the control group, calculated by weighted gene co-expression network analysis (WGCNA), and sketched out the basic framework on a modular map with 25 functional modules targeted by DHI. Furthermore, the effective therapeutic module (ETM), defined as the highest correlation value with the phenotype alteration (ΔSAQ-AF, the change in SAQ-AF at Day 30 from baseline) calculated by WGCNA, was identified in the population with the best effect (ΔSAQ-AF ≥ 40), which is related to anticoagulation and regulation of cholesterol metabolism. We assessed the modular flexibility of this ETM using the global topological D value based on Euclidean distance, which is correlated with phenotype alteration (r2: 0.8204, P = 0.019) by linear regression. Our study identified the anti-angina therapeutic module in the effective population treated by the multi-target drug. Modular methods facilitate the discovery of network pharmacological mechanisms and the advancement of precision medicine. (ClinicalTrials.gov identifier: NCT01681316).
- 1 September 2021
- Published by Springer Nature in Signal Transduction and Targeted Therapy
- Vol. 6 (1) , 1-10
Open AccessDetection of an anti-angina therapeutic module in the effective population treated by a multi-target drug Danhong injection: a randomized trial
It’s a challenge for detecting the therapeutic targets of a polypharmacological drug from variations in the responsed networks in the differentiated populations with complex diseases, as stable coronary heart disease. Here, in an adaptive, 31-center, randomized, double-blind trial involving 920 patients with moderate symptomatic stable angina treated by 14-day Danhong injection(DHI), a kind of polypharmacological drug with high quality control, or placebo (0.9% saline), with 76-day following-up, we firstly confirmed that DHI could increase the proportion of patients with clinically significant changes on angina-frequency assessed by Seattle Angina Questionnaire (ΔSAQ-AF ≥ 20) (12.78% at Day 30, 95% confidence interval [CI] 5.86–19.71%, P = 0.0003, 13.82% at Day 60, 95% CI 6.82–20.82%, P = 0.0001 and 8.95% at Day 90, 95% CI 2.06–15.85%, P = 0.01). We also found that there were no significant differences in new-onset major vascular events (P = 0.8502) and serious adverse events (P = 0.9105) between DHI and placebo. After performing the RNA sequencing in 62 selected patients, we developed a systemic modular approach to identify differentially expressed modules (DEMs) of DHI with the Zsummary value less than 0 compared with the control group, calculated by weighted gene co-expression network analysis (WGCNA), and sketched out the basic framework on a modular map with 25 functional modules targeted by DHI. Furthermore, the effective therapeutic module (ETM), defined as the highest correlation value with the phenotype alteration (ΔSAQ-AF, the change in SAQ-AF at Day 30 from baseline) calculated by WGCNA, was identified in the population with the best effect (ΔSAQ-AF ≥ 40), which is related to anticoagulation and regulation of cholesterol metabolism. We assessed the modular flexibility of this ETM using the global topological D value based on Euclidean distance, which is correlated with phenotype alteration (r2: 0.8204, P = 0.019) by linear regression. Our study identified the anti-angina therapeutic module in the effective population treated by the multi-target drug. Modular methods facilitate the discovery of network pharmacological mechanisms and the advancement of precision medicine. (ClinicalTrials.gov identifier: NCT01681316).