Semi-automated Hand Annotation for Single Cell and Spatial Datasets
Beta Prior to Release: If you chose to use, please check back regularly for updates and improvements. Version 1.0 will indicate stable release and contain links to publication or preprint.
Creating SAHA Inputs #Coming Soon
Troubleshooting #Coming Soon
Comparing Multiple SAHA Runs #Coming Soon
SAHA is a user-friendly package with simple meta-data level input resulting in easy-to-understand exploration of cell annotation in single cell and spatial RNA sequencing datasets. Using either marker gene or average experession dataframes, the user has the option to compare the similarity of their unannotated clusters to any annotations in the literature.
IMPORTANTLY: This approach differs from label-transfer and integration-based approaches in that it does not require a large Seurat, SCE, Anndata, or any common data type in which the computational burden of the pipeline scales with the size of an object. Most (if not all) data fed into SAHA will be able to run on a laptop without the need for high-RAM devices or supercomputers.
Whether you can run every single cell package ever released or just finished your first single cell vignette, SAHA offers the options to Create, Investigate, or Refine cell annotations based on:
- Self-similarity: likeness of clusters within your query data
- Marker-based: presence of markers in a known database
- Marker-free: expression of non-marker genes in your data
- ComplexHeatmap
- circlize
- dplyr
- methods
- ggrepel
- eulerr
- reshape2
- corrplot
- ggpubr
- ape
- vegan
- dendextend
You can install the development version of SAHA like so:
library(devtools)
devtools::install_github("neurogenetics/SAHA")#main package
devtools::install_github("neurogenetics/SAHAdata")#data package containing Allen Brain Cell Atlas db inputsUpon stable release, instructions for CRAN installation will be posted here.