Repository navigation
User Guide
You can start your analysis at https://bhasinlab.bmi.emory.edu/PediatricSC/.
When first visiting the PedSCAtlas site, a pop-up will appear allowing the user to choose a pediatric cancer dataset to analyze from those currently available.
Once the user presses Start Analysis, the chosen cancer type's corresponding datasets (sc/snRNA-seq, bulkRNA-seq) will load and the pop-up will disappear after the datasets are finished loading. In order to choose a different cancer type to analyze, the user must refresh the page to restart their session. The user can begin their analysis by expanding any of the module tabs located on the left side of the page.
Once a dataset is chosen, the PedSCAtlas main page appears. The user adjusts the parameters and inputs information in the modules on the left sidebar, and the corresponding plots and information is presented in the Plots and Messages boxes. In order to save the plots generated by the PedSCAtlas, the user can right click on their generated plot and click Save Image As...
The SingleCell module offers users a way to visualize and interact with the sc- snRNA-seq data of their chosen pediatric cancer type. The user can filter the disease subtypes they want to analyze by checking and unchecking the subtype boxes in the Choose Disease Type to Show: section of the module. With every module, the user must click the Press to Update Plot button to generate a plot with their selected options.
Then, they can choose a method to analyze or visualize the dataset using the Viewing Format section. If the "UMAP with Groups" option is chosen, the user has the option to color the cells on the UMAP plot by assigned cluster, cell type, disease type, and more. If one of the "Gene Expression" viewing formats is chosen, the user can enter a gene of their choice and the gene expression will be plotted in a violin plot, grouped by the "data aspect" chosen on the x-axis. If the "Disease Type" or "Cell Type" options are chosen, an extra comparison option will appear, allowing the user to perform a statistical comparison of means.
The user can also filter the cells based on gene expression when "UMAP with Gene Expression" is chosen, using the slider at the bottom. If the user wants to only view cells with high expression, they can increase the slider closer to 100. The "Pathway Enrichment" viewing options allow the user to plot enrichment of pre-determined HALLMARK and BIOCARTA pathways.
If the "Ribbon Plot with Multiple Pathways" option is chosen, the user can visualize the differential enrichment of multiple pathways chosen within the cell types present in the dataset. Once the pathways are selected and two groups are selected to compare, a ribbon plot is generated with colored ribbons for if the pathway is highly enriched in each group.
The ImmuneCell module allows users to plot gene expression in the adult healthy bone marrow dataset. The user inputs a gene and a feature plot is generated, showing gene expression on a UMAP. This is useful for examining if a potential biomarker is malignant cell specific in blood cancers, or if there is some expression in non-malignant immune and hematopoietic stem cells.
The Biomarkers module allows users to plot enrichment of pre-determined disease subtype biomarker sets for their chosen pediatric cancer type. The enrichment can be plotted on a UMAP or plotted on a violin plot, grouped by the same "Data Aspects" choices available in the SingleCell module.
When the user chooses a disease subtype in the Biomarker Set section, the genes for that biomarker set are printed in the Messages box on the main page. For example, the B-MPAL biomarker set consists of S100A16, ATF3, GALNT2, KCNK12, MTRN2L12, HBEGF, CDKN1A, IFRD1, PLIN2, CD81, NR4A1, DDIT3, RFLNB, CBX4, and UBE2S genes.
The BulkExpression module allows users to view gene expression or gene enrichment in the bulk expression dataset of their chosen pediatric cancer type. This module is currently not available for the RMS dataset. The user can choose to perform statistical comparisons in the bottom box to examine if there are significant differences in expression or enrichment between groups, with options for Wilcox T-tests and One-Way ANOVA tests between subtypes.
In addition, the user can choose to perform survival analysis with their inputs by selecting the "Survival" option in the View Exp or Survival dropdown. They can choose to examine the survival association with the gene expression or enrichment in all disease subtypes or one specific subtype.
The Prediction module allows users to input their own acute leukemia normalized bulk RNA data, then the atlas will predict the acute leukemia subtype according to the expression of the genes in the biomarker sets for the subtypes available in the Biomarker module. See the file, example_data.csv, for the correct formatting and the specific genes needed for the bulk RNA expression input. If all the leukemia biomarker genes are not present in the user's dataset, the atlas will estimate the values for each sample based on the median sample value across the input genes.
Once the user inputs their correctly formatted, normalized bulk RNA expression csv file, the atlas will estimate the subtypes for each sample present and output a table, shown in the "Table" box of the main page.
As an example, the steps below show how a user interested in Acute Myeloid Luekemia (AML) would use the PedSCAtlas to analyze the differences between AML and other subtypes, validated their proposed biomarker, FLT3, and examine the survival association of their gene.
By using the SingleCell module and choosing UMAP with Groups and Cell Type options, the user can visualize the AML malignant cell (blast) locations on a UMAP. From the generated plot, we can see that the AML blasts are overlapping with T/My MPAL and B-ALL blasts, whereas B/My MPAL and T-ALL blasts are clustered more separately.
By using the SingleCell module and choosing Violin with Gene Expression and Disease Type options, the user can visualize FLT3 expression in different acute leukemia subtypes and young adult healthy samples. The user can examine if FLT3 has high expression in AML blasts versus other cell types, and if it has high expression versus the young adult healthy bone marrow cells. We can see that AML blasts have higher expression of FLT3 than other non-blast cell types, but B/My MPAL, T/My MPAL, and B-ALL blasts also have high expression of FLT3. The user could conclude that FLT3 expression is high in AML blasts, but it is not specific to AML among the acute leukemias.
By using the ImmuneCell module and plotting their proposed AML biomarker, FLT3, in healthy bone marrow cell types, they can examine if their biomarker is blast-cell specific, or if it has some expression in healthy immune or stem cells. We can see that it has mostly low expression, with some moderate expression in HSCs and other progenitor cell types. However, most expression is below the 1 threshold level. From this, the user could conclude that their biomarker is mostly blast specific, with the exception of some expression in progenitor cell types.
By using the BulkExpression module and plotting their proposed AML biomarker, FLT3, expression in bulkRNA-seq data, they can further examine if their biomarker has high expression in pediatric AML samples and perform a statistical comparison of means. The user choose to plot with the Expression and Compare Against AML (with Wilcox T-test) options, and a boxplot is generated. On the boxplot we can see that overall B/My MPAL has the highest mean expression, and AML has lower bulk expression than all other acute leukemia subtypes (B/My MPAL, T/My MPAL, Pre-B ALL) present.
By using the BulkExpression module again and choosing the Survival, cutP, and AML subtype options, the user can examine the overall survival association of FLT3 expression. We can see that in the AML samples, higher expression of FLT3 is associated with increased hazard (HR=1.5, p=0.041).