Accelerating single-cell genomic analysis with GPUs

Nolet, Corey; A, Lal; Ilango, Rajesh; Dyer, Taurean; Rajiv, Movva,; Zedlewski, John; Israeli, Johnny

doi:10.1101/2022.05.26.493607

Cited by 9 publications

(13 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We then constructed a kNN graph (k=30) in the RSS space and clustered the data set using the Leiden algorithm38 (resolution=80). For both steps we used the GPU-accelerated RAPIDS implementation which is provided through scanpy 51,53 . For all cell type marker genes on a given level in the hierarchy, we computed the area under the receiver operating characteristic curve (AUROC) as well as the detection rate across clusters.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

An integrated transcriptomic cell atlas of human neural organoids

He,

Dony,

Fleck

et al. 2023

Preprint

View full text Add to dashboard Cite

Neural tissues generated from human pluripotent stem cells in vitro (known as neural organoids) are becoming useful tools to study human brain development, evolution and disease. The characterization of neural organoids using single-cell genomic methods has revealed a large diversity of neural cell types with molecular signatures similar to those observed in primary human brain tissue. However, it is unclear which domains of the human nervous system are covered by existing protocols. It is also difficult to quantitatively assess variation between protocols and the specific cell states in organoids as compared to primary counterparts. Single-cell transcriptome data from primary tissue and neural organoids derived with guided or un-guided approaches and under diverse conditions combined with large-scale integrative analyses make it now possible to address these challenges. Recent advances in computational methodology enable the generation of integrated atlases across many data sets. Here, we integrated 36 single-cell transcriptomics data sets spanning 26 protocols into one integrated human neural organoid cell atlas (HNOCA) totaling over 1.7 million cells. We harmonize cell type annotations by incorporating reference data sets from the developing human brain. By mapping to the developing human brain reference, we reveal which primary cell states have been generated in vitro, and which are under-represented. We further compare transcriptomic profiles of neuronal populations in organoids to their counterparts in the developing human brain. To support rapid organoid phenotyping and quantitative assessment of new protocols, we provide a programmatic interface to browse the atlas and query new data sets, and showcase the power of the atlas to annotate new query data sets and evaluate new organoid protocols. Taken together, the HNOCA will be useful to assess the fidelity of organoids, characterize perturbed and diseased states and facilitate protocol development in the future.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

An integrated transcriptomic cell atlas of human neural organoids

He,

Dony,

Fleck

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Batch correction was then performed using Harmony ( 16 ) based on 17 lineage markers (CD163, CD68, CD19, CD20, CD21, CD3e, CD8, CD4, CD45RO, ICOS, FOXP3, E-CAD, Pan-Cytokeratin, Beta-Catenin, CD31, CD34, SMA) in order to minimize batch effects between the tissue slides. The feature matrix returned by Harmony was then used for unsupervised clustering using Leiden algorithm ( 17 ) and the GPU-accelerated package, Rapids ( 18 ). Leiden resolution equals to 1-6 were tested and the resolution of 4 was chosen for manual cluster annotation as it presents a sufficient number of clusters that separate the cells into smaller groups (N = 70) differentiated by protein expression variability over the technical variability.…”

Section: Methodsmentioning

confidence: 99%

Compartmentalized spatial profiling of the tumor microenvironment in head and neck squamous cell carcinoma identifies immune checkpoint molecules and tumor necrosis factor receptor superfamily members as biomarkers of response to immunotherapy

et al. 2023

View full text Add to dashboard Cite

Mucosal head and neck squamous cell carcinoma (HNSCC) are the seventh most common cancer, with approximately 50% of patients living beyond 5 years. Immune checkpoint inhibitors (ICIs) have shown promising results in patients with recurrent or metastatic (R/M) disease, however, only a subset of patients benefit from immunotherapy. Studies have implicated the tumor microenvironment (TME) of HNSCC as a major factor in therapy response, highlighting the need to better understand the TME, particularly by spatially resolved means to determine cellular and molecular components. Here, we employed targeted spatial profiling of proteins on a cohort of pre-treatment tissues from patients with R/M disease to identify novel biomarkers of response within the tumor and stromal margins. By grouping patient outcome categories into response or non-response, we show that immune checkpoint molecules, including PD-L1, B7-H3, and VISTA, were differentially expressed. Patient responders possessed significantly higher tumor expression of PD-L1 and B7-H3, but lower expression of VISTA. Analysis of response subgroups by Response Evaluation Criteria in Solid Tumors (RECIST) criteria indicated that tumor necrosis factor receptor (TNFR) superfamily members including OX40L, CD27, 4-1BB, CD40, and CD95/Fas, were associated with immunotherapy outcome. OX40L expression in tumor regions was higher in patient-responders than those with progressive disease (PD), while other TNFR members, CD27 and CD95/Fas were lower expressed in patients with a partial response (PR) compared to those with PD. Furthermore, we found that high 4-1BB expression in the tumor compartment, but not in the stroma, was associated with better overall survival (OS) (HR= 0.28, p-adjusted= 0.040). Moreover, high CD40 expression in tumor regions (HR= 0.27, p-adjusted= 0.035), and high CD27 expression in the stroma (HR= 0.2, p-adjusted=0.032) were associated with better survival outcomes. Taken together, this study supports the role of immune checkpoint molecules and implicates the TNFR superfamily as key players in immunotherapy response in our cohort of HNSCC. Validation of these findings in a prospective study is required to determine the robustness of these tissue signatures.

show abstract

“…For example, the GPU-accelerated tool RAPIDS [60] builds on top of the Dask framework in order to scale its data preparation and model training steps across multiple GPUs and machines [23]. In this assisting capacity, Dask has made it possible to scale single-cell analysis pipelines to upwards of millions of cells [61]. In large-scale distributed environments, i.e.…”

Section: Scaling Computational Biology With Daskmentioning

confidence: 99%

Scalable transcriptomics analysis with Dask: applications in data science and machine learning

2022

View full text Add to dashboard Cite

Background Gene expression studies are an important tool in biological and biomedical research. The signal carried in expression profiles helps derive signatures for the prediction, diagnosis and prognosis of different diseases. Data science and specifically machine learning have many applications in gene expression analysis. However, as the dimensionality of genomics datasets grows, scalable solutions become necessary. Methods In this paper we review the main steps and bottlenecks in machine learning pipelines, as well as the main concepts behind scalable data science including those of concurrent and parallel programming. We discuss the benefits of the Dask framework and how it can be integrated with the Python scientific environment to perform data analysis in computational biology and bioinformatics. Results This review illustrates the role of Dask for boosting data science applications in different case studies. Detailed documentation and code on these procedures is made available at https://github.com/martaccmoreno/gexp-ml-dask. Conclusion By showing when and how Dask can be used in transcriptomics analysis, this review will serve as an entry point to help genomic data scientists develop more scalable data analysis procedures.

show abstract

Accelerating single-cell genomic analysis with GPUs

Cited by 9 publications

References 22 publications

An integrated transcriptomic cell atlas of human neural organoids

An integrated transcriptomic cell atlas of human neural organoids

Compartmentalized spatial profiling of the tumor microenvironment in head and neck squamous cell carcinoma identifies immune checkpoint molecules and tumor necrosis factor receptor superfamily members as biomarkers of response to immunotherapy

Scalable transcriptomics analysis with Dask: applications in data science and machine learning

Contact Info

Product

Resources

About