timothy-barry / ondisc Goto Github PK

Hi,

I tried to use "create_ondisc_matrix_from_mtx" function, but got the following error when writing CSR data:
Error in list2DF(.) : could not find function "list2DF"
In addition: Warning message:
In fun(libname, pkgname) :
Option 'datatable.old.unique.by.key' has been removed as warned for 4 years. It is now ignored. Please use by=key(DT) instead and stop using this option.

Version in my environment:
R version: 3.6.1, data.table version: 1.14.0

I attempted to create_ondisc_matrix_from_R_matrix with an ngCMatrix input (sparse logical matrix), and got the error message Error in ondisc::create_ondisc_matrix_from_R_matrix(r_matrix = gRNA_assignment_matrix, : Input matrix must be a class of matrix, dgTMatrix, dgCMatrix, or dgRMatrix.

• Replace readr with data.table globally.
• Wean off rhdf5; use C++ directly.
• Remove features_only initialization option.
• run_core_algo_step_1 computes n_nonzero_features for each chunk and saves as a list; this is somewhat memory inefficient and likely unnecessary. Just compute n_nonzero_features across ALL chunks.
• Replace %>% with |>

I tried passing a sparse matrix in dgCMatrix format to create_ondisc_matrix_from_R_matrix(), but it said that the input is required to be in dgRMatrix format (or perhaps it could also be in dense format). Could you please add support for dgCMatrix input?

Hi Tim,
I downloaded the processed import-gasperini-2019-v2 data from the dropbox link you shared in sceptre2-manuscript. I want to generated a sceptre object using the code in import-gasperini-2019-v2.

However, when I ran
gene_odm <- read_odm(paste0(processed_gene_dir, "matrix.odm"), paste0(processed_gene_dir, "metadata.rds"))

I got an error:
Error in read_odm(paste0(processed_gene_dir, "matrix.odm"), paste0(processed_gene_dir, : could not find function "read_odm"

Is the read_odm funtion has been deleted? How I can get the sceptre object from these input ?

>tree ./gasperini-2019-v2/
gasperini-2019-v2/
└── at-scale
    └── processed
        ├── gene
        │   ├── matrix.odm
        │   └── metadata.rds
        ├── grna_assignment
        │   ├── grna_assignment.zip
        │   ├── matrix.odm
        │   └── metadata.rds
        ├── grna_expression
        │   ├── grna_expression.zip
        │   ├── matrix.odm
        │   └── metadata.rds
        └── pairs_ungrouped.rds

5 directories, 9 files

Best
Yan

The create_ondisc_matrix_from_mtx function appears to use a lot of memory when the chunked option is set to true; this likely is due to a bug in read_delim_chunked. Consider different approaches to implementing this functionality (e.g., Linux split) if the bug is not fixed.

Note: this vignette takes a while to recompile. Talk to Tim about a shortcut.