Call NATMI Pipeline from R with Resources Querried from OmniPath [[DEPRECATED]]
Source:R/natmi_pipe.R
call_natmi.Rd
Call NATMI Pipeline from R with Resources Querried from OmniPath [[DEPRECATED]]
Usage
call_natmi(
sce,
op_resource,
expr_file = "em.csv",
meta_file = "metadata.csv",
output_dir = "NATMI_test",
reso_name = "placeholder",
assay = "RNA",
num_cor = 4,
conda_env = NULL,
assay.type = "logcounts",
.format = TRUE,
.overwrite_data = TRUE,
.seed = 1004,
.natmi_path = NULL,
.delete_input_output = FALSE
)
Arguments
- sce
Seurat or SingleCellExperiment object
- op_resource
List of OmniPath resources
- expr_file
expression matrix file name
- meta_file
annotations (i.e. clusters) file name
- output_dir
NATMI output directory
- reso_name
name of the resource usually in the format list(name = op_resource)
- assay
Seurat assay to be used
- num_cor
number of cores to be used
- conda_env
name of the conda environment via which NATMI is called
- assay.type
logcounts by default, but it's converted back into counts as suggested by the authors
- .format
bool whether to format output
- .overwrite_data
bool whether Extract and overwrite csv with data from Seurat Object
- .seed
random seed
- .natmi_path
path of NATMI code and dbs (by default set to liana path)
- .delete_input_output
logical whether to delete input and output after run.
Details
This function will save NATMI dbs folder, then it will call the NATMI Python from the NATMI dir and save the output into a specified directory in NATMI's path. It will then load the csvs and format the output to a list of lists.
By default, NATMI's path is set to that of LIANA, but any alternative path can be passed
============================================================================== NATMI Arguments: --interDB INTERDB lrc2p (default) has literature supported ligand-receptor pairs | lrc2a has putative and literature supported ligand-receptor pairs | the user-supplied interaction database can also be used by calling the name of database file without extension --interSpecies INTERSPECIES human (default) | mouse | expandp | expanda --emFile EMFILE the path to the normalised expression matrix file with row names (gene identifiers) and column names (cell-type/single-cell identifiers) --annFile ANNFILE the path to the metafile in which column one has single-cell identifiers and column two has corresponding cluster IDs (see file 'toy.sc.ann.txt' as an example) --species SPECIES human (default) | mouse | rat | zebrafish | fruitfly | chimpanzee | dog | monkey | cattle | chicken | frog | mosquito | nematode | thalecress | rice | riceblastfungus | bakeryeast | neurosporacrassa | fissionyeast | eremotheciumgossypii | kluyveromyceslactis, 21 species are supported --idType IDTYPE symbol (default) | entrez(https://www.ncbi.nlm.nih.gov/gene) | ensembl(https://www.ensembl.org/) | uniprot(https://www.uniprot.org/) | hgnc(https://www.genenames.org/) | mgi(http://www.informatics.jax.org/mgihome/nomen/index.shtml) | custom(gene identifier used in the expression matrix) --coreNum CORENUM the number of CPU cores used, default is one --out OUT the path to save the analysis results (Taken From NATMI's GitHub Page)
Stats: 1) The mean-expression edge weights 2) The specificity-based edge weights * a weight of 1 means both the ligand and receptor are only expressed in one cell type
Note that `call_natmi` will write the expression matrix to CSV each time its called, unless .overwrite_data is set to FALSE! This can be an extremely time consuming step when working with large datasets
Also, NATMI will sometimes create duplicate files, so please consider saving each run in a new folder. An easy fix would be to simply delete the output, but I am reluctant to automatically delete files via an R script.