In YosefLab/FastProjectR: Functional interpretation of single cell RNA-seq latent manifolds

Functional Interpretation for
scRNA-seq Data

VISION aids in the interpretation of single-cell RNA-seq (scRNA-seq) data by selecting for gene signatures which describe coordinated variation between cells. While the software only requires an expression matrix and a signature library (available in online databases), it is also designed to integrate into existing scRNA-seq analysis pipelines by taking advantage of precomputed dimensionality reductions, trajectory inferences or clustering results. The results of this analysis are made available through a dynamic web-app which can be shared with collaborators without requiring them to install any additional software.

• Nature Communications publication
• GitHub
• Biorxiv Preprint

Installing VISION

We recommend installing VISION via github using devtools:

library(devtools)
install_github("YosefLab/VISION")

See the DESCRIPTION file for a complete list of R dependencies. If the R dependencies are already installed, installation should finish in a few minutes.

The VISION Pipeline

VISION generally follows the same pipeline from iteration to iteration, where minor differences can be specified via the various parameters in a VISION object. On a typical VISION run:

• For large datasets, or if the user so chooses, micropools are computed - grouping similar cells together to reduce the complexity of the analysis.
• If a latent space is not specified, PCA is performed and the top 30 components are retained.
• A KNN graph is constructed from the latent space, named the cell-cell similarity map
• Signature scores are computed using the expression matrix
• Signature local “consistencies” on the cell-cell similarity map are computed using the Geary-C statistic, an auto-correlation statistic.
• An interactive web-based report is generated that can be used to explore and interpret the dataset.

How to run VISION

You can refer to the Getting Started page for instructions on running VISION. Additional vignettes are available in the Tutorials menu above.

Sample Output

• Live output example: ~9,000 CBMC's sequenced with the CITE-seq protocol
• Tour of the output report user interface(PDF)

YosefLab/FastProjectR documentation built on June 13, 2024, 4:20 p.m.