Filtered Dataset Import about metacell HOT 4 CLOSED

Original comment by Yaniv Lubling (Bitbucket: 557058:11761e0c-f009-41f9-a29c-6f6993e0539c, ).

Hi,

We do not support import of a SingleCellExpriment of Seurat objects. You can load several 10x datasets with the function mcell_import_multi_scmat_10x.

The dataset of origin of each cell is stored in the @cell_metadata slot of the mat object, together with any additional information you'll supply on the datasets in the dataset_table_fn parameter of mcell_import_multi_scmat_10x.

Yaniv

from metacell.

Original comment by Amos Tanay (Bitbucket: 557058:5a3503bb-91a6-4af0-972c-ab1d0c3bb50f, GitHub: amostanay).

If you are comfortable with R, then In principle it is also not hard to construct the metacell matrix object directly using:

mat=scm_new_matrix(my_matrix, my_metadata)
scdb_add_mat("new_mat", mat)

The matrix parameter should be a UMI table, and the metadata should be a data.frame with row for each cell, with mandatory columns: batch_set_id, amp_batch_id, seq_batch_id.

It is probably however better to go through the standard import and make sure all the metadata is well aligned. We will look into the possibility of adding more import features, including from Seurat objects.

Amos

from metacell.

Original comment by Michael Durante (Bitbucket: 5c3e72b636647b5b91027b5b, GitHub: mdurante1).

Thank you for providing this helpful information. I will try to both of these suggestion to import data to the proper metacell format. If I use the mcell_import_multi_scmat_10x function to import 10X data is it possible to subset the data based on given barcodes, clustering, or any other method? I am interested in looking at a subset of cells within my datasets. If not, I will attempt to use Amos' suggestion.

Thanks,
Michael

from metacell.

Original comment by Amos Tanay (Bitbucket: 557058:5a3503bb-91a6-4af0-972c-ab1d0c3bb50f, GitHub: amostanay).

Yes - subsetting a matrix object is easy using the scm_sub_mat command.

For example the following code generate a submatrix from all cells with metadata value "magic_field" at 2 or 6:

mat = scdb_mat("mymat")
metadata = mat@cell_metadata
mycells = rownames(metadata)[metadata$magic_field==2 | metadata$magic_field ==6 ]
mycells = intersect(mycells, colnames(mat@mat) #this is needed in case the matrix already ignored some of the cells, e.g. those with low UMI count
mat_to_use = scm_sub_mat(mat, cells=mycells)
scdb_add_mat("my_submat", mat_to_use)

This can be altered to facilitate any filtering of cells.

from metacell.