Output Files

Generating Output

A FastProjectOutput object can be generated using the Analyze function:

fpo <- Analyze(fp)

This single function will run the entirety of the analysis given the parameters of the FastProject object created (refer to How to Run FastProject on Your Data). Upon completing the analysis, one can view the output before saving the output object

viewResults(fpo)

or after saving the output object

saveFPOutAndViewResults(fpo)

Using the HTML Report

Sample Output Report

The HTML report currently supports a myriad of ways to interact with the results of the signatures vs projections plot. First and foremost is the table functionality to the left of the page. There are 10 tables, representing clusters of signatures based on their rank-sums across all cells. Each cell of the table is color-coded by the relative significance of the log_10 p-value associated with its consistency scores on that particular projection (which are organized by the table's columns). By clicking on the a signature name itself, the cluster will expand and display all signatures in that particular cluster. By selecting a cell in the table, the projection will display in the right hand visualization box where the points (cells) are colored by the signature's consistency scores or rank values for the given projection.

The box on the lower left hand side serves as a secondary information and visualization box, providing tabbed windows for alternative analyses. Firstly, a clustered heatmap can be generated from the signature vs. projection plot -- many different clustering methods can theoretically be applied, yet only Kmeans is supported currently. By selecting various numbers of clusters to generate, the heatmap displays the clusters appropriately; by hovering over a given cluster on the heatmap, the corresponding number of cells in the signature vs. projection plot will illuminate.

The second tab in the secondary visualization box provides the ability to view more informatin about the selected signature. In particular, the box will supply information regarding its file source, any meta data given with the signature file, a list of the genes in the signature and each gene's sign, and links to the Gene Cards of each gene. Clicking on the gene name will open an additional page with the gene card.

Finally, the third tab allows subset analysis of the cells. Frequently, a two dimensional projection will expose an interesting subset of cells that a user may wish to further analyze. The user can either click select, lasso select (by enabling Lasso Selection in the upper right hand side of the page), or range select cells (based on rank or consistency score values) to further analyze. A new FastProject object will automatically be created and sent off for analysis. When the analysis is done, the object will be saved and a new report will be displayed in a new tab. To note, this is especially important when pooling is applied to the input expression matrix. When the number of cells in an expression matrix is greater than 35,000, a clustering algorithm is applied to diminish the effective number of cells in the analysis (FastProjectR Pooling Algorithm). When viewing the output report, selecting points for further analysis will result in further analyzing larger clusters of cells on a more granular level.