Baohong Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Baohong Zhang has a total of 206 co-authors. Here are the top co-authors by number of shared publications:
- Kejie Li
Research Department, Biogen, Inc., 225 Binney St., Cambridge, MA 02142, USA
9 shared publications - Shanrong Zhao
Integrative Biology Center of Excellence, Pfizer Worldwide Research and Development, Cambridge, Massachusetts 02139, USA
9 shared publications - Yu Huining Sun
Department of Biology, University of Rochester, Rochester, USA
9 shared publications Zhengyu Ouyang
Data Science, BioInfoRx Inc. , Madison, WI 53719
9 shared publications- Wenxing Hu
Research Department, Biogen, Inc., 225 Binney St., Cambridge, MA 02142, USA
8 shared publications Jing Zhu
6 shared publications
Latest Publications
- Open Access
- 20 March 2026
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 42 (4)
SpliceHarmonization: an integrated method for identifying RNA splicing events in therapeutics for splicing modulation
Splicing, a critical co-transcriptional process in eukaryotes, enhances transcriptome diversity by generating isoforms specific to cell types, tissues, or developmental stages. Recent advancements in splicing modulators have opened new avenues for targeting previously undruggable genes by inducing significant perturbations in splicing events. These developments underscore the need for comprehensive methods to accurately identify and compare splicing events. While several tools have been developed to detect local splice variants, inconsistencies across methods remain a significant challenge. To address this, we present SpliceHarmonization, an integrated approach that combines the strengths of rMATS, LeafCutter, and MAJIQ, enabling robust and reliable splicing analysis with event type annotations. In a comprehensive evaluation using diverse simulated datasets, SpliceHarmonization streamlined and standardized the outputs from three detection methods into a unified format, thereby improving splicing detection with event type annotation and outperforming individual methods. By integrating the outputs from rMATS, LeafCutter and MAJIQ, our approach not only enhanced identification of a wide range of splicing events but also effectively mitigated method-specific discrepancies. This integration led to an accuracy exceeding 0.8 and a recall of up to 0.5, with an observed increase in AUC of up to 10%. Furthermore, SpliceHarmonization demonstrated high sensitivity in detecting low-abundance and complex splicing events, providing annotations including genomic coordinates and event type. SpliceHarmonization is available at https://github.com/interactivereport/SpliceHarmonization. Supplementary Materials and Supplementary Methods are available at Bioinformatics online. Supplementary Data is available at https://zenodo.org/records/15032402.
- 20 March 2026
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 42 (4)
Open AccessSpliceHarmonization: an integrated method for identifying RNA splicing events in therapeutics for splicing modulation
Splicing, a critical co-transcriptional process in eukaryotes, enhances transcriptome diversity by generating isoforms specific to cell types, tissues, or developmental stages. Recent advancements in splicing modulators have opened new avenues for targeting previously undruggable genes by inducing significant perturbations in splicing events. These developments underscore the need for comprehensive methods to accurately identify and compare splicing events. While several tools have been developed to detect local splice variants, inconsistencies across methods remain a significant challenge. To address this, we present SpliceHarmonization, an integrated approach that combines the strengths of rMATS, LeafCutter, and MAJIQ, enabling robust and reliable splicing analysis with event type annotations. In a comprehensive evaluation using diverse simulated datasets, SpliceHarmonization streamlined and standardized the outputs from three detection methods into a unified format, thereby improving splicing detection with event type annotation and outperforming individual methods. By integrating the outputs from rMATS, LeafCutter and MAJIQ, our approach not only enhanced identification of a wide range of splicing events but also effectively mitigated method-specific discrepancies. This integration led to an accuracy exceeding 0.8 and a recall of up to 0.5, with an observed increase in AUC of up to 10%. Furthermore, SpliceHarmonization demonstrated high sensitivity in detecting low-abundance and complex splicing events, providing annotations including genomic coordinates and event type. SpliceHarmonization is available at https://github.com/interactivereport/SpliceHarmonization. Supplementary Materials and Supplementary Methods are available at Bioinformatics online. Supplementary Data is available at https://zenodo.org/records/15032402.
- Open Access
SpaceSequest: A unified pipeline for spatial transcriptomics data analysis
Background: Spatial transcriptomics has emerged as one of the most powerful tools for gaining biological insights, enabling researchers to uncover intricate relationships between gene expression patterns and tissue architecture. Recent advances in the field have resulted in a variety of new platforms, including Visium, Visium HD, and Xenium from 10x Genomics, as well as GeoMx and CosMx from NanoString Technologies, which has now been acquired by the Bruker Corporation. However, the existence of diverse spatial transcriptomics platforms and various data formats poses challenges in standardizing data analysis. Thus, there remains a critical gap in the availability of a comprehensive pipeline capable of conducting end-to-end analysis that is necessary to extract biological insights from multiple spatial transcriptomics platforms. Results: Here, we present SpaceSequest, a tailored pipeline that utilizes cutting-edge computational methodologies to conduct a thorough analysis, enabling the extraction of crucial biological insights from five major spatial transcriptomics technologies. SpaceSequest performs (1) standardized quality control and general data processing, (2) key analyses customized for each spatial platform, (3) automated cell type annotation and deconvolution, and (4) high-quality figure and analysis result generation. In addition, SpaceSequest allows for smooth integration with cellxgene VIP and Quickomics for user-friendly data access and interactive visualization. Conclusions: SpaceSequest is a unified and comprehensive pipeline designed for the analysis, visualization, and publication of spatial transcriptomics data from various platforms. The source code is available at https://github.com/interactivereport/SpaceSequest . To facilitate seamless installation and usage, we have also created a detailed Bookdown tutorial that can be accessed through https://interactivereport.github.io/SpaceSequest/tutorial/docs/index.html .
Open AccessSpaceSequest: A unified pipeline for spatial transcriptomics data analysis
Background: Spatial transcriptomics has emerged as one of the most powerful tools for gaining biological insights, enabling researchers to uncover intricate relationships between gene expression patterns and tissue architecture. Recent advances in the field have resulted in a variety of new platforms, including Visium, Visium HD, and Xenium from 10x Genomics, as well as GeoMx and CosMx from NanoString Technologies, which has now been acquired by the Bruker Corporation. However, the existence of diverse spatial transcriptomics platforms and various data formats poses challenges in standardizing data analysis. Thus, there remains a critical gap in the availability of a comprehensive pipeline capable of conducting end-to-end analysis that is necessary to extract biological insights from multiple spatial transcriptomics platforms. Results: Here, we present SpaceSequest, a tailored pipeline that utilizes cutting-edge computational methodologies to conduct a thorough analysis, enabling the extraction of crucial biological insights from five major spatial transcriptomics technologies. SpaceSequest performs (1) standardized quality control and general data processing, (2) key analyses customized for each spatial platform, (3) automated cell type annotation and deconvolution, and (4) high-quality figure and analysis result generation. In addition, SpaceSequest allows for smooth integration with cellxgene VIP and Quickomics for user-friendly data access and interactive visualization. Conclusions: SpaceSequest is a unified and comprehensive pipeline designed for the analysis, visualization, and publication of spatial transcriptomics data from various platforms. The source code is available at https://github.com/interactivereport/SpaceSequest . To facilitate seamless installation and usage, we have also created a detailed Bookdown tutorial that can be accessed through https://interactivereport.github.io/SpaceSequest/tutorial/docs/index.html .
- Open Access
- 15 September 2025
- Published by Springer Nature in Nature Methods
- Vol. 22 (9) , 1911-1922
Scaling up spatial transcriptomics for large-sized tissues: uncovering cellular-level tissue architecture beyond conventional platforms with iSCALE
Recent advances in spatial transcriptomics (ST) technologies have transformed our ability to profile gene expression while preserving crucial spatial context within tissues. However, existing ST platforms are constrained by high costs, long turnaround times, low resolution, limited gene coverage and inherently small tissue capture areas, which hinder their broad applications. Here we present iSCALE, a method that reconstructs large-scale, super-resolution gene expression landscapes and automatically annotates cellular-level tissue architecture in samples exceeding capture areas of current ST platforms. The performance of iSCALE was assessed by comprehensive evaluations involving benchmarking experiments, immunohistochemistry staining and manual annotations by pathologists. When applied to multiple sclerosis human brain samples, iSCALE uncovered lesion-associated cellular characteristics undetectable by conventional ST experiments. Our results demonstrate the utility of iSCALE in analyzing large tissues by enabling unbiased annotation, resolving cell type composition, mapping cellular microenvironments and revealing spatial features beyond the reach of standard ST analysis or routine histopathological assessment.
- 15 September 2025
- Published by Springer Nature in Nature Methods
- Vol. 22 (9) , 1911-1922
Open AccessScaling up spatial transcriptomics for large-sized tissues: uncovering cellular-level tissue architecture beyond conventional platforms with iSCALE
Recent advances in spatial transcriptomics (ST) technologies have transformed our ability to profile gene expression while preserving crucial spatial context within tissues. However, existing ST platforms are constrained by high costs, long turnaround times, low resolution, limited gene coverage and inherently small tissue capture areas, which hinder their broad applications. Here we present iSCALE, a method that reconstructs large-scale, super-resolution gene expression landscapes and automatically annotates cellular-level tissue architecture in samples exceeding capture areas of current ST platforms. The performance of iSCALE was assessed by comprehensive evaluations involving benchmarking experiments, immunohistochemistry staining and manual annotations by pathologists. When applied to multiple sclerosis human brain samples, iSCALE uncovered lesion-associated cellular characteristics undetectable by conventional ST experiments. Our results demonstrate the utility of iSCALE in analyzing large tissues by enabling unbiased annotation, resolving cell type composition, mapping cellular microenvironments and revealing spatial features beyond the reach of standard ST analysis or routine histopathological assessment.
- Open Access
CompBioAgent : An LLM-powered agent for single-cell RNA-seq data exploration
A bstract: Advancements in high-throughput biological technologies, particularly single-cell RNA sequencing (scRNA-seq), have generated vast amounts of complex data that offer valuable insights into gene expression, cellular behavior, and disease mechanisms. However, the analysis of such data often requires specialized computational skills and expertise, which can pose a barrier to many researchers. To address this challenge, we developed CompBioAgent, a user-friendly web application designed to democratize access to bioinformatics resources, powered by Large Language Models (LLMs). By integrating with the CellDepot, CompBioAgent allows users to easily query and explore gene expression data related to various diseases, cell types, and experimental conditions. Moreover, the tool employs Cellxgene Visualization In Plugin (Cellxgene VIP) platform to generate a range of intuitive visualizations, including violin plots, UMAP embeddings, heatmaps, etc. With its natural language interface, CompBioAgent automatically converts scientific queries into structured data requests and returns the plot of interest, enabling seamless exploration of biological data without the need for programming expertise. Such integration enhances the ability of researchers to visualize complex biological insights and identify key patterns, making high-level data analysis more accessible and efficient. A demo website and a list of examples are available at https://celldepot.bxgenomics.com/compbioagent , and the source code is released at https://github.com/interactivereport/compbioagent .
Open AccessCompBioAgent : An LLM-powered agent for single-cell RNA-seq data exploration
A bstract: Advancements in high-throughput biological technologies, particularly single-cell RNA sequencing (scRNA-seq), have generated vast amounts of complex data that offer valuable insights into gene expression, cellular behavior, and disease mechanisms. However, the analysis of such data often requires specialized computational skills and expertise, which can pose a barrier to many researchers. To address this challenge, we developed CompBioAgent, a user-friendly web application designed to democratize access to bioinformatics resources, powered by Large Language Models (LLMs). By integrating with the CellDepot, CompBioAgent allows users to easily query and explore gene expression data related to various diseases, cell types, and experimental conditions. Moreover, the tool employs Cellxgene Visualization In Plugin (Cellxgene VIP) platform to generate a range of intuitive visualizations, including violin plots, UMAP embeddings, heatmaps, etc. With its natural language interface, CompBioAgent automatically converts scientific queries into structured data requests and returns the plot of interest, enabling seamless exploration of biological data without the need for programming expertise. Such integration enhances the ability of researchers to visualize complex biological insights and identify key patterns, making high-level data analysis more accessible and efficient. A demo website and a list of examples are available at https://celldepot.bxgenomics.com/compbioagent , and the source code is released at https://github.com/interactivereport/compbioagent .
- Open Access
Scaling up spatial transcriptomics for large-sized tissues: uncovering cellular-level tissue architecture beyond conventional platforms with iSCALE
Recent advances in spatial transcriptomics (ST) technologies have transformed our ability to profile gene expression while retaining the crucial spatial context within tissues. However, existing ST platforms suffer from high costs, long turnaround times, low resolution, limited gene coverage, and small tissue capture areas, which hinder their broad applications. Here we present iSCALE, a method that predicts super-resolution gene expression and automatically annotates cellular-level tissue architecture for large-sized tissues that exceed the capture areas of standard ST platforms. The accuracy of iSCALE were validated by comprehensive evaluations, involving benchmarking experiments, immunohistochemistry staining, and manual annotation by pathologists. When applied to multiple sclerosis human brain samples, iSCALE uncovered lesion associated cellular characteristics that were undetectable by conventional ST experiments. Our results demonstrate iSCALE’s utility in analyzing large-sized tissues with automatic and unbiased tissue annotation, inferring cell type composition, and pinpointing regions of interest for features not discernible through human visual assessment.
Open AccessScaling up spatial transcriptomics for large-sized tissues: uncovering cellular-level tissue architecture beyond conventional platforms with iSCALE
Recent advances in spatial transcriptomics (ST) technologies have transformed our ability to profile gene expression while retaining the crucial spatial context within tissues. However, existing ST platforms suffer from high costs, long turnaround times, low resolution, limited gene coverage, and small tissue capture areas, which hinder their broad applications. Here we present iSCALE, a method that predicts super-resolution gene expression and automatically annotates cellular-level tissue architecture for large-sized tissues that exceed the capture areas of standard ST platforms. The accuracy of iSCALE were validated by comprehensive evaluations, involving benchmarking experiments, immunohistochemistry staining, and manual annotation by pathologists. When applied to multiple sclerosis human brain samples, iSCALE uncovered lesion associated cellular characteristics that were undetectable by conventional ST experiments. Our results demonstrate iSCALE’s utility in analyzing large-sized tissues with automatic and unbiased tissue annotation, inferring cell type composition, and pinpointing regions of interest for features not discernible through human visual assessment.
- Open Access
Spatial transcriptomics reveals heterogeneous cell–cell interactions among brain regions in a cuprizone model consistent with multiple sclerosis lesions
The cuprizone (CPZ) model is widely used for modeling demyelination in multiple sclerosis (MS) and for testing potential remyelination therapies. We integrated single-cell and spatial transcriptomics (ST) to fine map the spatial cellular and molecular responses during de and remyelination. ST revealed global demyelination and neuroinflammation in the brain beyond the corpus callosum, with region-specific differences. We identified oligodendroglia and microglia as two major cell types with significant transcriptomic changes in the model. Ligand receptor pairing analyses predicted growth factor and phagocytic pathway enrichment during demyelination, which is consistent with changes in MS lesions. During remyelination, while mature oligodendrocytes nearly reversed their phenotype back to the control state, microglia remained associated with the demyelination phenotype. Finally, astrocytes in the CPZ model had the greatest preservation of disease-associated modules to MS lesions, while the MOL, OPC, and microglia showed moderate to low preservation, which overall suggested that the CPZ model had moderate translatability to chronically active MS lesions.
Open AccessSpatial transcriptomics reveals heterogeneous cell–cell interactions among brain regions in a cuprizone model consistent with multiple sclerosis lesions
The cuprizone (CPZ) model is widely used for modeling demyelination in multiple sclerosis (MS) and for testing potential remyelination therapies. We integrated single-cell and spatial transcriptomics (ST) to fine map the spatial cellular and molecular responses during de and remyelination. ST revealed global demyelination and neuroinflammation in the brain beyond the corpus callosum, with region-specific differences. We identified oligodendroglia and microglia as two major cell types with significant transcriptomic changes in the model. Ligand receptor pairing analyses predicted growth factor and phagocytic pathway enrichment during demyelination, which is consistent with changes in MS lesions. During remyelination, while mature oligodendrocytes nearly reversed their phenotype back to the control state, microglia remained associated with the demyelination phenotype. Finally, astrocytes in the CPZ model had the greatest preservation of disease-associated modules to MS lesions, while the MOL, OPC, and microglia showed moderate to low preservation, which overall suggested that the CPZ model had moderate translatability to chronically active MS lesions.
- Open Access
xOmicsShiny: an R shiny application for cross-omics data analysis and pathway mapping
Summary: We developed xOmicsShiny, a feature-rich R Shiny-powered application that enables biologists to fully explore omics datasets across experiments and data types, with an emphasis on uncovering biological insights at the pathway level. The data merging feature ensures flexible exploration of cross-omics data, such as transcriptomics, proteomics, metabolomics, and lipidomics. The pathway mapping function covers a broad range of databases, including WikiPathways, Reactome, and KEGG pathways. In addition, xOmicsShiny offers several visualization options and analytical tasks for everyday omics data analysis, namely, PCA, Volcano plot, Venn Diagram, Heatmap, WGCNA, and advanced clustering analyses. The application employs customizable modules to perform various tasks, generating both interactive plots and publication-ready figures. This dynamic, modular design overcomes the issue of slow loading in R Shiny tools and allows it to be readily expanded by the research and developer community. Availability and implementation: The R Shiny application is publicly available at: http://xOmicsShiny.bxgenomics.com . Researchers can upload their own data to the server or use the pre-loaded demo dataset. The source code, under MIT license is provided at https://github.com/interactivereport/xOmicsShiny for local installation. A full tutorial of the application is available at https://interactivereport.github.io/xOmicsShiny/tutorial/docs/index.html . Contact: yuhenry.sun@biogen.com or baohong.zhang@biogen.com Supplementary data: Supplementary data are available at bioRxiv online.
Open AccessxOmicsShiny: an R shiny application for cross-omics data analysis and pathway mapping
Summary: We developed xOmicsShiny, a feature-rich R Shiny-powered application that enables biologists to fully explore omics datasets across experiments and data types, with an emphasis on uncovering biological insights at the pathway level. The data merging feature ensures flexible exploration of cross-omics data, such as transcriptomics, proteomics, metabolomics, and lipidomics. The pathway mapping function covers a broad range of databases, including WikiPathways, Reactome, and KEGG pathways. In addition, xOmicsShiny offers several visualization options and analytical tasks for everyday omics data analysis, namely, PCA, Volcano plot, Venn Diagram, Heatmap, WGCNA, and advanced clustering analyses. The application employs customizable modules to perform various tasks, generating both interactive plots and publication-ready figures. This dynamic, modular design overcomes the issue of slow loading in R Shiny tools and allows it to be readily expanded by the research and developer community. Availability and implementation: The R Shiny application is publicly available at: http://xOmicsShiny.bxgenomics.com . Researchers can upload their own data to the server or use the pre-loaded demo dataset. The source code, under MIT license is provided at https://github.com/interactivereport/xOmicsShiny for local installation. A full tutorial of the application is available at https://interactivereport.github.io/xOmicsShiny/tutorial/docs/index.html . Contact: yuhenry.sun@biogen.com or baohong.zhang@biogen.com Supplementary data: Supplementary data are available at bioRxiv online.
- Open Access
ScaleSC: A superfast and scalable single cell RNA-seq data analysis pipeline powered by GPU
The rise of large-scale single-cell RNA-seq data has introduced challenges in data processing due to its slow speed. Leveraging advancements in GPU computing ecosystems, such as CuPy, and building on Scanpy and rapids-singlecell package, we developed ScaleSC, a GPU-accelerated solution for single-cell data processing. ScaleSC delivers over a 20x speedup through GPU computing and significantly improves scalability, handling datasets of 10–40 million cells with over 1000 batches by overcoming the memory bottleneck on a single A100 card- far surpassing rapids-singlecell’s capacity of processing only 1 million cells without multi-GPU support. We also resolved discrepancies between GPU and CPU algorithm implementations to ensure consistency. In addition to core optimizations, we developed new advanced tools for marker gene identification, cluster merging, and more, with GPU-optimized implementations seamlessly integrated. Designed for ease of use, the ScaleSC package is compatible with Scanpy workflows, requiring minimal adaptation from users. The ScaleSC package ( https://github.com/interactivereport/ScaleSC ) promises significant benefits for the single-cell RNA-seq computational community.
Open AccessScaleSC: A superfast and scalable single cell RNA-seq data analysis pipeline powered by GPU
The rise of large-scale single-cell RNA-seq data has introduced challenges in data processing due to its slow speed. Leveraging advancements in GPU computing ecosystems, such as CuPy, and building on Scanpy and rapids-singlecell package, we developed ScaleSC, a GPU-accelerated solution for single-cell data processing. ScaleSC delivers over a 20x speedup through GPU computing and significantly improves scalability, handling datasets of 10–40 million cells with over 1000 batches by overcoming the memory bottleneck on a single A100 card- far surpassing rapids-singlecell’s capacity of processing only 1 million cells without multi-GPU support. We also resolved discrepancies between GPU and CPU algorithm implementations to ensure consistency. In addition to core optimizations, we developed new advanced tools for marker gene identification, cluster merging, and more, with GPU-optimized implementations seamlessly integrated. Designed for ease of use, the ScaleSC package is compatible with Scanpy workflows, requiring minimal adaptation from users. The ScaleSC package ( https://github.com/interactivereport/ScaleSC ) promises significant benefits for the single-cell RNA-seq computational community.
- Open Access
- 13 January 2025
- Published by MDPI AG in Bioengineering
- Vol. 12 (1) , 57
FundusNet: A Deep-Learning Approach for Fast Diagnosis of Neurodegenerative and Eye Diseases Using Fundus Images
Early-stage detection of neurodegenerative diseases is crucial for effective clinical treatment. However, current diagnostic methods are expensive and time-consuming. In this study, we present FundusNet, a deep-learning model trained on fundus images, for rapid and cost-effective diagnosis of neurodegenerative diseases. FundusNet achieved superior performance in age prediction (MAE 2.55 year), gender classification (AUC 0.98), and neurodegenerative disease diagnosis—Parkinson’s disease AUC 0.75 ± 0.03, multiple sclerosis AUC 0.77 ± 0.02. Grad-CAM was used to identify which part of the image contributes to diagnosis. Imaging biomarker interpretation demonstrated that FundusNet effectively identifies clinical retina structures associated with diseases. Additionally, the model’s high accuracy in predicting genetic risk suggests that its performance could be further enhanced with larger training datasets.
- 13 January 2025
- Published by MDPI AG in Bioengineering
- Vol. 12 (1) , 57
Open AccessFundusNet: A Deep-Learning Approach for Fast Diagnosis of Neurodegenerative and Eye Diseases Using Fundus Images
Early-stage detection of neurodegenerative diseases is crucial for effective clinical treatment. However, current diagnostic methods are expensive and time-consuming. In this study, we present FundusNet, a deep-learning model trained on fundus images, for rapid and cost-effective diagnosis of neurodegenerative diseases. FundusNet achieved superior performance in age prediction (MAE 2.55 year), gender classification (AUC 0.98), and neurodegenerative disease diagnosis—Parkinson’s disease AUC 0.75 ± 0.03, multiple sclerosis AUC 0.77 ± 0.02. Grad-CAM was used to identify which part of the image contributes to diagnosis. Imaging biomarker interpretation demonstrated that FundusNet effectively identifies clinical retina structures associated with diseases. Additionally, the model’s high accuracy in predicting genetic risk suggests that its performance could be further enhanced with larger training datasets.
- Open Access
- 26 December 2024
- Published by Oxford University Press (OUP) in Bioinformatics Advances
- Vol. 5 (1) , vbaf167
ScaleSC: a superfast and scalable single-cell RNA-seq data analysis pipeline powered by GPU
Summary: The rise of large-scale single-cell RNA-seq data has introduced challenges in data processing due to its slow speed. Leveraging advancements in Graphics Processing Unit (GPU) computing ecosystems, such as CuPy and Compute Unified Device Architecture (CUDA), building on Scanpy and Rapids-singlecell package, we developed ScaleSC, a GPU-accelerated solution for large-scale single-cell data processing. ScaleSC delivers over a 20× speedup through GPU computing and significantly improves scalability, handling datasets of 10–20 million cells with over 1000 batches by overcoming the memory bottleneck on a single A100 card, which far surpasses Rapids-singlecell’s capacity of processing only 1 million cells without multi-GPU support. We also resolved discrepancies between GPU and Central Processing Unit (CPU) algorithm implementations to ensure consistency. In addition to core optimizations, we developed novel tools for marker gene identification and cluster merging with GPU-optimized implementations seamlessly integrated. ScaleSC shares a similar syntax with Scanpy, which helps lower the learning curve for users already familiar with Scanpy workflows. Availability and implementation: The ScaleSC package (https://github.com/interactivereport/ScaleSC) promises significant benefits for the single-cell RNA-seq computational community.
- 26 December 2024
- Published by Oxford University Press (OUP) in Bioinformatics Advances
- Vol. 5 (1) , vbaf167
Open AccessScaleSC: a superfast and scalable single-cell RNA-seq data analysis pipeline powered by GPU
Summary: The rise of large-scale single-cell RNA-seq data has introduced challenges in data processing due to its slow speed. Leveraging advancements in Graphics Processing Unit (GPU) computing ecosystems, such as CuPy and Compute Unified Device Architecture (CUDA), building on Scanpy and Rapids-singlecell package, we developed ScaleSC, a GPU-accelerated solution for large-scale single-cell data processing. ScaleSC delivers over a 20× speedup through GPU computing and significantly improves scalability, handling datasets of 10–20 million cells with over 1000 batches by overcoming the memory bottleneck on a single A100 card, which far surpasses Rapids-singlecell’s capacity of processing only 1 million cells without multi-GPU support. We also resolved discrepancies between GPU and Central Processing Unit (CPU) algorithm implementations to ensure consistency. In addition to core optimizations, we developed novel tools for marker gene identification and cluster merging with GPU-optimized implementations seamlessly integrated. ScaleSC shares a similar syntax with Scanpy, which helps lower the learning curve for users already familiar with Scanpy workflows. Availability and implementation: The ScaleSC package (https://github.com/interactivereport/ScaleSC) promises significant benefits for the single-cell RNA-seq computational community.