Skip to content

Latest commit

 

History

History
16 lines (10 loc) · 1.37 KB

README.md

File metadata and controls

16 lines (10 loc) · 1.37 KB
Raw

Scripts for merging datasets

Scripts in this directory were used for merging processed datasets. The final merged gene expression matrix is too large to be represented as a sparse matrix in R, so cannot be represented as a Seurat object. To merge all datasets:

  1. merge_datasets.R This will create two Seurat objects with the standardized metadata and assays and save the result to h5seurat files. Some additional standardization occurs in this step:

    • Gene names are standardized, using the most current mappings from the HUGO Gene nomenclature committee. This mapping table is provided in this directory as hgncAlias2Symbol.tsv
    • Genes are filtered to only include those present in all datasets. For simplicity, this list of genes is provided in genes.txt.

  2. merge_sce.R
    This merges the two Seurat objects created from merge_datasets.R utilizing the Bioconductor SingleCellExperiment object and HDF5Array, which allows matrices to be referenced from disk without reading into memory, and saves the result as a SingleCellExperiment object with the transcriptomic data saved in hdf5 format.

  3. merge_sct.py
    This script merges the two Seurat objects created from merge_datasets.R utilizing the Python hdf5 API to coerce the merged object into the h5seurat format, which can be transformed into the web portal.